 Current DPD on running book of Green Malabar is 4.

97% while they claim that NPA

till may 31st is 3.2%. we didn’t receive the flow data for vivrity.
 Their process of underwriting and disbursement don’t have any external influence
but they have done out of policy (below 680 and SMA’s customers) with MD
influence
 As on Mar'23, 90+ book is 3.13CR (4.79% in AUM terms) and 2.19% from
disbursement standpoint
 2.35CR is above 180+ DPD.
 For May'19-Dec'19 cohort the static pool NPA is significantly high (>9%). In general
the old cohorts not getting stabilized
 90+ book went up to 3.13CR compare to 1.9CR in Mar'22
 They claim for high recovery rate but that is not reflected in data.
 In the Dec-Feb cohort shows some good recovery and that they claim that this is
started after the management change in Dec but again the portfolio recovery burst in
mar.

 Thanks everyone,

 Now let me convey the complete scenarios from last year,

 We discussed and closed the pricing + MSA with geospoc last year. Further-on we
had a discussion with KD regarding the integration and under guidance of KD it was
concluded that that as it is risk consumable service so risk should own it. Hence
Paresh has done the integration on risk server. PO is already raised for Geospoc and
we are production ready for risk system consumption.
 we have standard practice that we don’t use any third-party service directly without
our own intelligence so we have a model on top of it to define the negative area but
that we will use only when we will have pan India presence of collection and right
now we are going with collection negative area list (which you must have from
collection to implement).
 About the second service (I.e. address correction) they have we took market
feedback, We didn’t get good response so we stopped exploring the same.

 So here few learning we can have:

 Any new third party vendor communication for Pay per use service offerings: If any
vendor connect to anyone, please add Prashant at very first place (This is the
standard practice every team is doing) with expected usecase to explore and identify
 Any confirmation about vendor communication: Prashant is the SPOC for the same,
you can connect Prashant and he may guide you the status of vendor and further
spoc from different team to get the usage/integration level.
 Keep the discussion in documented form (All of us need to do this) to avoid any
Chinese whisper

 Now regarding the discussion with Geospoc, as they have already engaged with us so
I don’t see any point to have a further discussion from risk side. Furtheron you are
 free to take on any new services they are offering which are not covered under
current MSA.

 Team,
 Thanks everyone,

 The dealings and arrangements with Geospoc from last year are as follows:

 The pricing and Master Service Agreement (MSA) with Geospoc were finalized.
Afterward, integration guidance was sought from KD. KD affirmed that the risk team
should manage the risk consumable service. As a result, Paresh completed the
integration on our risk server. The Purchase Order (PO) for Geospoc has been issued,
and we are set for production on the risk system.

 We have an internal protocol (For risk science team) in place that prohibits the direct
use of any third-party service without our proprietary intelligence layer. A model to
identify negative areas has been added on top of the Geospoc service. However, it
will only be activated upon the establishment of a pan-India collection presence or a
direct request from collection team. For now, rely on the collection negative area list
that you must have already obtained from the collection team.

 As for Geospoc's second service, address correction, market feedback has been
gathered. The response was not favourable, so further exploration was halted.

 Below are some essential rules:

 For any new third-party vendor communication involving Pay-per-use service

offerings: If a vendor reaches out, add Prashant immediately and detail the
anticipated use case for exploration and identification.
 Regarding vendor communication confirmation: Prashant is the designated Single
Point of Contact (SPOC). Reach out to him for the status of any vendor and other
teams' SPOCs for usage or integration levels.
 Keep all discussions documented to prevent any miscommunication.

 Regarding Geospoc, further discussions from the risk side are pointless, as they are
already engaged with us. Although You are free to explore any new services they
offer that are not covered under the current MSA.

 I applied for medical insurance policy with policybazar which is going to be provided
by Care insurance. But today when I enquired about the status they said that they
require my medical hospitalization certificate and disability certificate as in the
declaration form I mentioned that I have polio.
 Now the problem is:

 disabled is itself a wrong word and question on the dignity of a person. and just
because I have polio, labelling me disabled is a serious criminal offense. In case
someone want to call out govt declared word is differently abled/physically
 challenged. Calling someone in any circumstance a disabled person is a serious
criminal offense and violation to the fundamental rights
 Disability certificate is not a medical certificate and should not be asked to a
customer. Insurer asked this certificate because of lazy underwriting and creating
unnecessary issue.
 As I declared myself that I am fully fit, not on any medication, never hospitalized and
never have any major or minor issue, that should be treated as sole source of truth. If
insurer have any doubt they can send medical inspection team to check my health
situation

 So I explained this to the customer care representative (Nancy and her immediate
senior). As a response they didn’t even admit that insurer should be confronted here.
On the place of the same they were continuously asking that either I should provide
disability certificate, or declare that I don’t have or say that I’ll not provide (Further
on policy will be cancelled and I’ll get the full refund). I even asked that connect me
to the insurer so that I’ll ask the same to them directly and reasoning for the same
they denied that they are not allowed as per policy.
 Now my issue is
 what is the role of policy bazar in overall deal (Just a messaging agent to pass the
communication directly from insurer to the customer and say that either you will
provide the meaningless ask or you will not get the policy)
 If policybazar people don’t understand the gravity of matter why they don’t involve
seniors or make sure customers would have a direct discussion with the insurer.

 Hereby I would like to confirm that I never ever have any hospitalization due to polio,
no single major medical checkup nor I have created a disability certificate. So as the
following action I would like to expect that:

 Here onwards no labeling of disabled to me

 Faster process of policy decision
 Apology email for labeling me as disabled just because I have polio

 Dear Policy Bazaar Team,

 I hope this email finds you well. I am writing to express my grave concerns and deep
dissatisfaction regarding the recent process related to my application for a medical
insurance policy via your platform, to be provided by Care Insurance.

 I disclosed my polio condition honestly in the declaration form, and it was met with
demands for my medical hospitalization certificate and a so-called 'disability
certificate'. I would like to bring to your attention three major issues with these
requirements:

 The term 'disabled' is in itself offensive and degrading, deeply disrespectful to the
dignity of a person. Using such language is not only a serious affront but also a legal
offense. We ought to use dignified language like 'differently-abled' or 'physically
 challenged'. I urge you to revise your policies and ensure that all employees
understand this vital aspect.
 A disability certificate is not synonymous with a medical certificate and must not be
demanded from customers. The insistence on this document appears to stem from
careless underwriting practices, leading to unnecessary complications and undue
stress for clients.
 I have declared myself fully fit, not on any medication, never hospitalized, and
without any major or minor health issues. This statement should be acknowledged
and treated as the sole source of truth. If there are doubts, an appropriate course of
action would be to send a medical inspection team to verify my health situation.

 When I discussed these issues with your representatives, Nancy and her immediate
senior, they failed to provide satisfactory assistance. Their insistence on obtaining a
disability certificate or else canceling the policy, coupled with their refusal to connect
me with the insurer, is deeply troubling. It calls into question Policy Bazaar's role in
this transaction and its commitment to customer service.

 In light of the above, I request the following actions:

 Stop referring to me as 'disabled' immediately.

 Expedite the policy decision process.
 Issue a formal apology for using inappropriate language.

 In summary, I have not been hospitalized due to my polio condition, have never had
a major medical checkup, nor have I ever issued a disability certificate. I strongly urge
you to consider my words and rectify the situation. Failure to do so will leave me no
option but to consider other legal and consumer protection avenues.
 I look forward to a swift resolution to this issue. Please handle this matter with the
seriousness it deserves.

 My applied booking ID is: 565119130

 Dear Sir/Madam,

 I am Dhirendra Jha (current equity shareholders of Vinati organics Ltd.).

 Recently I started the process of dematization of my shares, the shareholding papers
was lost in transit.
 Now when I contacted my bank (SBI), the representative informed me that I need to
get duplicate of my shares issued from Vinati organics Ltd.

 So to help me with the same please reissue my physical shares. Please find below my
details.

 Name: Dhirendra Jha and Sona Jha

 Folio Number: D00764
 Distinctive No(s): 841511-843010
 Address: Dhirendra Jha
 o Sona Vila, Maulaganj, Brahmstahan, Darbhanga, Bihar- 846004

 Subject: Request for Reissuance of Physical Share Papers

 Dear Sir/Madam,

 I hope this email finds you well. I am writing to bring to your attention an issue I
encountered during the dematerialization process of my shares as an equity
shareholder of Vinati Organics Ltd. Unfortunately, the shareholding papers were lost
in transit. Upon contacting my bank, SBI, their representative advised me to obtain
duplicate share papers from Vinati Organics Ltd.

 I kindly request your assistance in reissuing my physical shares. Below are the
relevant details:

 Name: Dhirendra Jha and Sona Jha

 Folio Number: D00764
 Distinctive No(s): 841511-843010
 Address: Dhirendra Jha, Sona Vila, Maulaganj, Brahmstahan, Darbhanga, Bihar-
846004

 I would greatly appreciate it if you could initiate the necessary steps to issue the
duplicate share papers. Should you require any additional information or
documentation to facilitate this process, please do not hesitate to let me know.

 Thank you for your attention to this matter. I look forward to your prompt response
and the resolution of this issue. Please feel free to reach me at 9939734478 should
you require any further clarification or details.

 Yours sincerely,

 Dhirendra Jha

 Objective of this document is to understand the RFP requirement, proposed

solutions, deliverables

 SNAPSHOT
 I hope this correspondence finds you well. I am writing in response to the Request for
Proposal (RFP) that was graciously extended to Arthmate on the 29th of May, 2023, via
electronic mail. We express our gratitude for considering our organization as a potential
Anchor for the proposed project.
 We would like to acknowledge that we have duly received the RFP and have meticulously
reviewed its contents. As per your request, we have diligently compiled and included all the
necessary details in the accompanying document, which we believe will provide you with a
comprehensive understanding of our capabilities, expertise, and proposed solutions.
 At Arthmate, we pride ourselves on our commitment to delivering exceptional services and
products that cater to the unique requirements of our clients. As such, we have dedicated
significant time and resources to ensure that our response accurately reflects our
capabilities and aligns with the objectives outlined in the RFP.
 Furthermore, we have taken great care to address each section of the RFP thoroughly,
emphasizing our understanding of your project's intricacies and tailoring our proposed
solutions to meet your specific needs. Our team of experienced professionals has
collaborated extensively to devise a strategic and comprehensive approach that we believe
will exceed your expectations and deliver optimal results.
 Proposed Solution:
o We have carefully analyzed and summarized the requirements across different
product types. This comprehensive overview allows us to gain a holistic
understanding of the specific needs and preferences of our esteemed clients.
o The objective of our underwriting models and Business Rules Engine (BRE) system is
to develop an algorithm that can assess the creditworthiness of customers seeking a
loan in a fully automated manner. To achieve this, we will leverage data from
various sources to train the model. This data includes credit scores, payment history,
outstanding debts, income, and fraud-related information, which are crucial in
evaluating creditworthiness.
o Model Development process flow: The figure below illustrates the process flow to
be followed for the development of our model. This systematic approach ensures
that our underwriting model is robust and comprehensive, taking into account
various factors that influence creditworthiness. By leveraging advanced algorithms
and utilizing extensive data, we aim to provide an automated and
o Confidential
o accurate assessment of customer creditworthiness, streamlining the loan approval
process and enhancing efficiency.
o Model Validation and Benchmarking: Our model validation and benchmarking
framework encompasses the following key aspects:
o Validation Across Samples and Timeframes: The models will be rigorously validated
using different samples and timeframes, including in-sample and most recent data.
This comprehensive approach ensures the stability and reliability of the models over
time.
o Performance and Stability Metrics: Model performance and stability will be
evaluated using key metrics such as GINI, KS, and Capture rate. These metrics
provide a robust assessment of the model's predictive power, discriminatory ability,
and ability to capture target outcomes accurately.
o Benchmarking Analysis: Benchmarking analysis will be conducted by comparing our
newly developed models with traditional scores. The objective is to measure the
impact on approval rates and default rates. Our aim is to demonstrate that the
newly developed model generates higher approval rates while maintaining
significantly low default rates, showcasing the superiority of our approach.
 The figure below provides a sample benchmarking analysis:
o 8

o Confidential
o By implementing this model validation and benchmarking framework, we ensure
that our models are stable, perform well across various samples and timeframes,
 and outperform traditional scoring systems. This rigorous approach allows us to
deliver accurate and reliable credit assessments while maximizing approvals and
minimizing default rates.
o Model Deployment flow: The model deployment architecture depicted in the figure
below showcases on-premise deployment of all models. However, it's important to
note that the actual deployment plan may be subject to variation based on the
availability of infrastructure. Figure below exhibits proposed deployment
architecture.
o In this architecture, the models are deployed on local infrastructure, ensuring that
the entire deployment process occurs within the organization's premises. This
approach provides control over data privacy and security.
o The specifics of the infrastructure can vary based on the organization's requirements
and available resources. It could involve dedicated servers, clusters, or a
combination of hardware and virtualized environments. The infrastructure should be
capable of handling the computational requirements of the deployed models
efficiently.
o Additionally, the deployment plan may include components such as load balancers,
firewalls, and monitoring tools to ensure optimal performance, security, and
scalability. These components help manage incoming requests, protect against
unauthorized access, and provide insights into the system's health and performance.
o The actual deployment plan should be designed based on careful assessment of the
organization's needs, the complexity of the models, and the scalability
requirements. It is crucial to involve experts in infrastructure management and
deployment to ensure a robust and efficient deployment process.
o Remember, the figure provided is an illustration of the model deployment
architecture, and the actual deployment plan may vary depending on the
infrastructure availability and specific organizational requirements.
o 9

o Confidential
o Realtime Underwriting process Flow: After the completion of model development,
testing, and deployment, a seamless integrated system will be established to
facilitate the real-time customer journey for availing a loan. This system will
consume various services such as underwriting models and business rules engines
(BRE) to generate the following outputs:
o Customer Accept/Reject/Refer: The system will determine whether a customer's
loan application should be accepted, rejected, or referred for further evaluation. It
will leverage the underwriting models and other relevant data to make an informed
decision based on predefined criteria and risk assessment.
o Customer eligible Loan Amount: Using the underwriting models and borrower's
financial information, the system will calculate the maximum loan amount that the
customer is eligible to borrow. This assessment will consider factors such as income,
credit score, debt- to-income ratio, and other relevant criteria.
o Customer Risk-based Interest rate: Based on the risk assessment conducted by the
underwriting models and other relevant factors, the system will determine the
appropriate interest rate for the customer. The interest rate will be tailored to the
individual's risk profile, ensuring a fair and customized pricing structure.
 By integrating these services into the loan availing process, the system will
enable a real-time customer journey that provides quick and accurate
 decisions on loan acceptance, determines the eligible loan amount, and
offers risk-based interest rates. This streamlined process enhances the
customer experience, improves operational efficiency, and facilitates
informed lending decisions.
 Figure below exhibit sample flow diagram of end-to-end flow of real time
decisioning process.
o 10

o Collection Models:
o The objective of collection models in the lending business is to proactively manage
and minimize missed payments, enabling an effective collection strategy. This is
achieved through the following two types of models:
o Confidential
o Early Delinquency Control: The first type of model is designed to predict the
likelihood of EMI bounce in the upcoming month for customers who are currently
non-delinquent. By training this model using a development sample consisting of
non-delinquent customers, it helps identify cases with a high probability of missing
their EMI early on. This allows for targeted interventions and proactive measures to
prevent delinquency, ensuring timely payments and reducing the number of
accounts that fall into delinquency.
o Late Delinquency Control: The second type of model focuses on predicting the
likelihood of customers rolling forward into higher delinquency buckets, specifically
targeting customers who are already 30+ days past due (dpd). By training this model
using a development sample of customers in the 30+ dpd category, it enables early
identification of cases likely to slip further into higher dpd buckets. This knowledge
facilitates an efficient collection strategy, allowing prompt actions to be taken to
minimize slippage and ensure timely recoveries.
 By developing and leveraging these collection models, the lending business
can implement a model- driven approach to collections. This approach
maximizes the effectiveness of the collection efforts, reduces collection
costs, and minimizes the number of accounts that progress into higher
delinquency buckets. Ultimately, the objective is to enhance overall
collections, improve cash flow, and maintain a healthy loan portfolio.
 Model Development and deployment process flow will remain almost same
as mentioned in underwriting model.
 Collection Usage Flow: An automated system will be developed to generate
the Collection score and corresponding collection strategy 5-7 days before
the EMI cycle. Each customer will be assigned one of four priorities, ranging
from Priority 1 (the riskiest segment) to Priority 4 (the best segment), based
on the desired action. This prioritization allows for efficient allocation of
resources, ensuring minimal
o The automated system will utilize the Collection score to determine the appropriate
strategy for each customer, streamlining the collection process and optimizing
outcomes. By assigning priorities and implementing tailored strategies, the system
ensures effective and targeted efforts are employed, reducing collection costs and
maximizing the chances of successful EMI collection.
o Figure below exhibit sample flow diagram of running collection process.
o Confidential
o Cross-sell and Upsell Models:
o The objective of cross-selling and upselling models is to provide customers with
suitable products at the right time, thereby preventing them from seeking services
from other lenders. The cross-selling model aims to offer complementary or related
products to customers based on their needs and preferences. By identifying
additional products that enhance the customer's experience or fulfill their additional
requirements, the model ensures that the customer remains engaged with the same
lender rather than seeking alternatives.
o On the other hand, the upselling model focuses on identifying opportunities to
provide top-up options on existing products for underleveraged customers. These
customers may have the potential to benefit from additional features or higher-
value versions of the product they already possess. By leveraging the upselling
model, lenders can tailor their offers to these customers, encouraging them to
upgrade or avail top-up options that align with their financial goals.
o The cross-selling and upselling models work together to enhance customer
satisfaction and loyalty while maximizing the lender's revenue. By offering suitable
products and top-up options, the models increase the chances of retaining
customers and ensuring that they continue their financial journey with the same
lender. This approach also helps to build stronger customer relationships and
reduces the risk of losing customers to competitors offering similar services.
o Model Development: To develop cross-selling and upselling models, a combination
of existing loan repayment data and industry data will be leveraged. These datasets
provide valuable insights into customer behavior, preferences, and trends, which
can be used to identify potential cross-selling and upselling opportunities.
o One aspect of developing these models involves creating a propensity model, which
predicts the likelihood of a customer to take a particular loan or product. By
analyzing historical data and customer attributes, the model identifies patterns and
factors that contribute to a customer's interest in specific offerings. This information
allows lenders to understand which products are more likely to resonate with each
customer and tailor their recommendations accordingly.
o In addition to propensity modeling, probability of default models are utilized to
assess the risk associated with offering a particular product to a customer. These
models analyze various factors
o Confidential
o 12
o such as credit history, income, and financial indicators to estimate the probability of
a customer defaulting on the loan. By incorporating this information, lenders can
make informed decisions about which products are suitable for each customer,
considering their risk profile and minimizing the potential for defaults.
o Once these models are developed, they enable lenders to identify and display
relevant products to customers through various channels, such as online platforms
or personalized marketing campaigns. By utilizing the insights gained from the
propensity and probability of default models, lenders can make targeted offers to
customers, increasing the likelihood of successful cross-selling and upselling.
o By leveraging existing loan repayment and industry data, along with the utilization of
propensity models and probability of default models, lenders can effectively identify
the right products to present to each customer. This data-driven approach enhances
 the customer experience by offering personalized recommendations while also
managing the risk associated with offering new products.
o Model Usage Flow Diagram: To run the cross-selling program, follow these steps:
o Cross-engine Generation: On a specific periodic basis, the cross-selling engine will
generate suitable products based on customer data and preferences. These
products are identified as potential cross-selling opportunities.
o Automated System: The system will run in an automated fashion, presenting the
suitable product to customers through various channels, such as online platforms or
targeted marketing campaigns. The system utilizes data analysis and algorithms to
match the right product with the right customer.
o Customer Interest: If a customer shows interest in the offered product, they will be
redirected to the normal underwriting journey. This involves the standard
procedures and assessments required for loan approval, including verifying
eligibility, evaluating creditworthiness, and conducting risk assessments.
o Separate Business Rules Engine (BRE): A separate business rules engine will be
designed specifically for the cross-selling and upselling program. This BRE will
incorporate rules and criteria specific to cross-selling and upselling initiatives,
ensuring that the right products are recommended to customers at the right time.
 By following this process, the cross-selling program operates systematically
and efficiently. The automated system presents suitable products to
customers, and if they express interest, they proceed through the standard
underwriting journey. The separate BRE for cross-selling and upselling
ensures that the program's specific rules and criteria are implemented
effectively.
 Figure below exhibits the process flow of running the cross-sell engine.
o Confidential

o Ongoing Support:
o For ongoing monitoring and maintenance of models and systems, we offer a
comprehensive 12- month support package. This package includes continuous
monitoring of all deployed models and systems to ensure their optimal
performance.
o As part of the monitoring process, we conduct periodic assessments to identify any
potential issues or discrepancies. If any issues are detected, the models undergo a
thorough review and analysis. In cases where significant issues are found, the
models will be redeveloped to address the identified issues and improve their
performance.
o Our goal is to ensure the models and systems remain accurate, efficient, and aligned
with changing business requirements throughout the maintenance period. By
providing regular monitoring and proactive maintenance, we help to uphold the
reliability and effectiveness of the models and systems over time.
o Portfolio Monitoring Dashboard:
o The proposed portfolio monitoring dashboard in the lending business aims to
provide a comprehensive and visual representation of the portfolio. It offers the
following features:
o Holistic Portfolio View: The dashboard captures a holistic view of the portfolio,
allowing stakeholders to gain insights and monitor various aspects of lending
operations.
o Data Warehouse Integration: The dashboard is linked to a data warehouse, enabling
users to access charts and data with just a click of a button. This integration ensures
real-time and up- to-date information.
o Sourcing Mix Analysis: The dashboard presents the sourcing mix across different
dimensions such as Score band, Geography, Income band, Occupation type, and
other key dimensions defined by the Business Rules Engine (BRE). This analysis
provides a clear understanding of the customer acquisition patterns.
o Coincidental Delinquencies and Disbursed Amount: It showcases coincidental
delinquencies, disbursed amount, and outstanding amount across different
segments, allowing for effective monitoring of delinquency levels and loan
performance.
o Confidential
o 14
o Static Pool/Vintage Curves: The dashboard includes static pool/vintage curves,
which help in understanding the delinquency buildup over time. This analysis aids in
assessing the performance of different cohorts or batches of loans.
o Bucket-wise Collection Efficiency: Users can view bucket-wise collection efficiency
across segments, providing insights into the effectiveness of collection efforts for
different stages of delinquency.
o Actual vs Expected Credit Loss: The dashboard displays actual vs expected credit
loss across segments, enabling comparison and analysis of the portfolio's credit risk
performance.
o On-Us and Off-Us Delinquency Charts: The dashboard features charts that illustrate
the delinquency patterns of customers in relation to repayment behavior with both
the lender ("On-Us") and competitors ("Off-Us"). This analysis helps identify
customer repayment trends and potential competitive challenges.
 By offering a comprehensive and visually intuitive representation of the
portfolio, the proposed dashboard enhances portfolio monitoring and
facilitates informed decision-making in the lending business.
 Sample dashboard view is attached here.
 Dashboard_vf.pptx

 Technical capabilities:
 Arthmate Group's technical capabilities include:
o Skilled Team: The company has a team of 170+ employees located in Kolkata,
Gurgaon, Mumbai, and Bangalore. The senior management team brings extensive
experience in Financial Services, Technology, and Consulting from renowned
organizations like ICICI Bank, American Express, BCG, Flipkart, Reliance, ITC, among
others.
o Digital-first Approach: Arthmate Group follows a digital-first approach in its work
processes, fostering collaboration and creating a low-waste, low-paper environment.
This approach focuses on enhancing efficiency and productivity.
o Tech and Data Science Expertise: Over 70% of the workforce comprises
professionals from the Tech and Data Science domains. This expertise enables the
company to leverage advanced technologies and data-driven insights in its
operations.
o Embedded Fintech Platform: Arthmate Group has developed an embedded fintech
platform that serves as a plug-n-play API-based lending stack. The platform includes
essential components such as Loan Origination System, Loan Management System,
and Multiple Underwriting Scorecard Models (A-Score, B-Score, and E-Score).
Additionally, the platform offers access to Digital India-stack services like Bureau,
cKYC, PAN, and more.
 o Integration Simplicity: Through the plug-n-play technology, Anchors can seamlessly
integrate with Arthmate Group and initiate lending activities within a short time-
frame of just three working days. Whether a Anchor requires all or specific services,
they can conveniently consume the company's APIs to fulfil their lending needs.
o 15

 Objective of this overall project is to digitize the overall underwriting process for
quara finance wherein following objective need to fulfil:

 Quara Finance is seeking a comprehensive Business Rule Engine and Credit Risk
Module for their credit underwriting system
 The solution should seamlessly integrate with their existing systems
 System should support data from both English and Arabic language
 System should be able to support both mobile app and web based journey

 Arthmatetech India Pvt Ltd (Tech and landing company based in India) shared the
RFP response to resolve the challenge as:

 Seek a comprehensive Business Rule Engine and Credit Risk Module for the credit
underwriting system.
 Ensure seamless integration of the solution with Quara Finance's existing systems.
 Support data from both English and Arabic languages to cater to a diverse customer
base.
 Enable the system to support both mobile app and web-based journeys.
 Customize the system to allow Quara Finance to define and manage their specific
business rules.
 Execute business rules in real-time to streamline the underwriting process.
 Automate credit risk assessment using advanced algorithms and data analytics
techniques.
 Provide a responsive dashboard to portfolio review and scorecard monitoring.
 Offer ongoing technical support, maintenance, and regular updates

 Our approach as response to these challenges are as follows:

 Scorecard based underwriting system to predict the probability of default

 Configurable BRE
 Month on month collection scorecard generation at user level

 Quara Finance aims to digitize its underwriting process with the following objectives:
 Underwriting and Collection Scorecards: Implement robust scorecards for advanced
risk assessment and optimized collection strategies.
 Customization: Build a flexible system for Quara Finance to define and manage
business rule within the scorecards.
 Integration: Develop a comprehensive Business Rule Engine and Credit Risk Module
that seamlessly integrates with Quara Finance's existing systems.
 Multilingual Support: Design the system to accommodate both English and Arabic
languages.
 Portfolio Monitoring Dashboard: Develop a comprehensive dashboard for real-time
insights into application statuses, underwriting, and collection performance, aiding
informed decision-making.

 Quara Finance aims to digitize its underwriting process with the following objectives:

 Scorecards: Implement robust scorecards for advanced risk assessment and

optimized collections.
 Customization: Build a flexible system for Quara Finance to define and manage
business rules.
 Integration: Develop a comprehensive Business Rule Engine and Credit Risk Module
that seamlessly integrates with Quara Finance's systems.
 Multilingual Support: Design the system to accommodate both English and Arabic
languages.
 Portfolio Monitoring: Develop a comprehensive dashboard with insights into
application statuses, underwriting, and collections.

 We will have 6 stage approach for the complete digitization of the process:


we will have 6 stage digitization approach:

 Assessment: Understand the current lending system, data points, and KSA regulatory
guidelines.
 Model Development: Create underwriting, collection, upsell, and cross-sell
models/systems and portfolio monitoring dashboard.
 Validation: Ensure models meet statistical parameters and obtain stakeholder signoff.
 Deployment: Set up UAT environment and integrate with Quara Finance system.
 Go Live Support: Assist Quara Finance's technical team in production deployment
and successful system launch.
 Maintenance: Provide continuous support for system operation and conduct regular
model monitoring to ensure optimal performance.

 For the success measurement of the project will have certain deliverables:

 Model development: Total 21 statistical model will be developed and delivered which
are as follows:
  B
us  Cr
in os
 Pr  Appl es s/  Coll
od icati
s Up ect
uc on
R Sel ion
t Scor
ul l Mo
Ty ecar
e Mo del
pe d
En de
gi l
ne
 Retail
 Mi
cr
o
 Y  Ye
Fi  Yes  Yes
es s
na
nc
e
 Ca
sh  Y  Ye
 Yes  Yes
Lo es s
an
 BN  Y  Ye
 Yes  Yes
PL es s
 SME
 M
ur
ab
ha  Y  Ye
 Yes  No
Fi es s
na
nc
e
 Po
int
 Y  Ye
of  Yes  No
es s
Sa
le
 Ija  Yes  Y  Ye  No
ra es s
 h

 Dashboards: Total 3 portfolio dashboard will be created i.e.

 Collection dashboard
 Cross-sell/Upsell dashboard
 Overall portfolio monitoring dashboard
 APIs: All the model and system consumption will be triggered through APIs. These
APis will have 3 types:
 Ingestion APIs: To ingest the data to the model for processing in predefined format.
 Consumption APIs: To consumed the processed output through system for real time
decisioning and analysis
 Webhook/callback: These are the notification APIs will be used to notify if the data is
processed and ready to consume
 Model Development: We will develop and deliver a total of 21 statistical models for
different product types, including application scorecards, business rule engines,
cross/up-sell models, and collection models. The models will cover various retail and
SME offerings such as microfinance, cash loans, BNPL, murabha finance, point of
sale, and ijarah.
 Dashboards: Three portfolio dashboards will be created to provide comprehensive
monitoring and insights. These include a collection dashboard, cross-sell/upsell
dashboard, and an overall portfolio monitoring dashboard.
 APIs: The system will be designed to utilize APIs for seamless integration and efficient
data processing. The APIs will include ingestion APIs to process data in a predefined
format, consumption APIs for real-time decisioning and analysis, and
webhook/callback APIs for notification when data is processed and ready for
consumption.

 The project deliverables include:

 Model Development:
o A total of 21 statistical models will be developed and delivered, covering
various product types.
o For each product type, the following models will be developed:
 Application Scorecard
 Business Rule Engine
 Cross/Up Sell Model
 Collection Model

 Dashboards:
o Three portfolio dashboards will be created:
 Collection Dashboard
 Cross-sell/Upsell Dashboard
 Overall Portfolio Monitoring Dashboard
 APIs: The models and systems will be accessed and utilized through APIs, which will
be categorized into three types:
o Ingestion APIs: To input data into the models for processing in a predefined
format.
 o Consumption APIs: To retrieve and utilize the processed output from the
system for real-time decision-making and analysis.
o Webhook/Callback APIs: Notification APIs that will be used to alert users
when the data is processed and ready for consumption.
 These deliverables aim to enhance the underwriting process, provide valuable
insights through dashboards, and enable seamless integration and utilization of the
models and systems through APIs.

 Application scorecards: These will be the product specific predictive model

developed for the product demography combination. Data fuel for these scorecard
will be the bureau data of customers fetched from any of the bureau in KSA. These
scorecard will be utilized to identify the probability of default on pre-decided bad
rate definition This helps to identify:
 Acceptance/Rejection
 Adequate risk premium (Risk adjusted pricing)
 Loan amount based on tradelines
 FOIR (If income is available)
o All above results to lower delinquency and higher approval rate
 BRE system: These are descriptive model helps to create go-nogo scenarios (based on
predefined rules including scorecard cutoff). These rules are defined and system will
be designed in such a way that what-if scenarios (impact of changing the rules) can
be seen in real time. This helps to perform:
 Acceptance/Rejection
 Scenario analysis
o Above results to transparent decisions and right cut off implementations
 Collection model: These are post disbursal model to predict customer’s probability of
not to pay categorises the customers in 3 categories based on the risk associated
with the customer. These processing are done at 5-10 days before to the EMI due
date so that collection strategy can be implemented accordingly. This helps to
identify:
 Upcoming delinquency
 Right collection strategy
o Above results to low NPA, with lesser collection cost
 Cross-sell and Upsell models: These are predictive models helps to encash more
opportunity through providing next based offer interms of same product with new
loan and next best product based on propensity to pay. At the same time
organization underwriting cut off can be implemented and a very focus set of
customers with most likely requirement is identified in lending. This provides us to
identify:
 Propensity to buy
 Propensity to pay
o Above results to create more opportunity for the organization with negligible
cost of conversion
 Portfolio dashboard: These are monitoring dashboard to understand customer pool
status, drop offs, scorecard performance and interactive scenarios for collection
agent with proper information what should be talked. This results as:
o Portfolio monitoring of slippage
 o Agent guiding system
o Opportunity identification

 Application Scorecards:
o Introduction: These product-specific predictive models leverage demographic
data and bureau information to assess the probability of default for a given
product-demographic combination.
o Benefits:
 Facilitate acceptance/rejection decisions.
 Enable risk-adjusted pricing by determining adequate risk premiums.
 Determine appropriate loan amounts based on tradelines.
 Calculate FOIR (if income is available).
 Result in lower delinquency rates and higher approval rates.
 BRE System:
o Introduction: The Business Rule Engine (BRE) system consists of descriptive
models that enable the creation of go-nogo scenarios based on predefined
rules, including scorecard cutoffs. Real-time what-if analysis allows for
dynamic assessment of the impact of rule changes.
o Benefits:
 Support acceptance/rejection decisions.
 Enable scenario analysis for decision-making.
 Ensure transparency in decisions and implementation of correct
cutoffs.
 Collection Model:
o Introduction: The collection model is a post-disbursal predictive model
designed to predict a customer's probability of not making payments. It
categorizes customers into three risk-based categories to implement
appropriate collection strategies.
o Benefits:
 Identify upcoming delinquency to take proactive measures.
 Implement the right collection strategy based on customer risk profile.
 Result in lower NPAs and reduced collection costs.
 Cross-sell and Upsell Models:
o Introduction: These predictive models identify opportunities for cross-selling
and upselling by offering the next-best product or loan based on a customer's
propensity to pay. Underwriting cutoffs can be implemented to target
customers with the highest likelihood of requiring additional financing.
o Benefits:
 Identify customers' propensity to buy and pay.
 Create more opportunities for the organization with minimal
conversion costs.
 Portfolio Dashboard:
o Introduction: The portfolio dashboard provides monitoring capabilities for
customer pool status, drop-offs, scorecard performance, and interactive
scenarios for collection agents. It facilitates informed decision-making and
improves efficiency.
 o Benefits:
 Monitor portfolio performance and identify potential slippage.
 Guide collection agents with relevant information for effective
communication.
 Identify new opportunities for the organization.

 Application Scorecards:
o Introduction: Predictive models for product-specific assessments, utilizing
demographic and bureau data to determine default probabilities.
o Benefits: Acceptance/rejection decisions, risk-adjusted pricing, loan amount
determination, FOIR calculation, lower delinquency, higher approval rates.
 BRE System:
o Introduction: Descriptive models for go-nogo scenarios, allowing real-time
analysis of rule changes.
o Benefits: Acceptance/rejection decisions, scenario analysis, transparent
decisions, correct cutoff implementation.
 Collection Model:
o Introduction: Post-disbursal model predicting non-payment probability,
categorizing customers for tailored collection strategies.
o Benefits: Identify delinquency, implement effective collection strategies, low
NPA, reduced collection costs.
 Cross-sell and Upsell Models:
o Introduction: Predictive models for targeted offers based on propensity to
pay, enabling cross-selling and upselling opportunities.
o Benefits: Propensity to buy and pay, increased opportunities, minimal
conversion costs.
 Portfolio Dashboard:
o Introduction: Monitoring dashboard for customer pool status, scorecard
performance, and interactive collection scenarios.
o Benefits: Portfolio monitoring, agent guidance, opportunity identification.

Quara Finance's underwriting digitization project objectives:

 Scorecards: Implement robust risk assessment, cross-sell, and optimized collections.

 Customization: Flexible system for defining and managing business rules.
 Integration: Seamless integration of Business Rule Engine and Credit Risk Module
with existing systems.
 Multilingual Support: English and Arabic language compatibility.
 Portfolio Monitoring: Comprehensive dashboard for application statuses,
underwriting, cross-sell, and collections insights.

Arthmate's 6-stage digitization approach:

  Assessment: Understand current system, data, and regulatory guidelines.
 Model Development: Create underwriting, collection, upsell, cross-sell models, and
portfolio monitoring dashboard.
 Validation: Ensure models meet statistical parameters and obtain stakeholder signoff.
 Deployment: Set up UAT environment and integrate with Quara Finance system.
 Go Live Support: Assist Quara Finance in production deployment and system launch.
 Maintenance: Provide continuous support and conduct regular model monitoring for
operational efficiency.

The Project deliverables include:

Model Development:
 Development of 21 statistical models covering various product types (Appendix 1)
 Models for each product type: Application Scorecard with dynamic pricing and EMI
estimation, Business Rule Engine, Cross/Up Sell Model, Collection Model

Dashboards:
 Creation of three portfolio dashboards:
 Collection Dashboard
 Cross-sell/Upsell Dashboard
 Overall Portfolio Monitoring Dashboard
APIs:
 Implementation of APIs for accessing and utilizing the models and systems,
categorized into three types:
 Ingestion APIs: Input data into the models in a predefined format
 Consumption APIs: Retrieve and utilize processed output for real-time
decision-making and analysis
 Webhook/Callback APIs: Notification system to alert users when data is
processed

Integration and Maintenance:

 Integration and maintenance support for the developed model, dashboard, and
system by Arthmate.

I am Rahul Kumar, Vice president and Head of risk for Arthmate. I have total 12+ years of
experience in Analytics, Strategy and RISK.

By education, I am a passedout of IIm Calcutta and computer science engineer from BIT
Sindri.

My professional experience roaming around various domains wherein I started my career as

head of analytics for ecommerce organization Naaptol. Followed on I moved to prestigious
tata group as a strategist and worked with telecom domain. I further worked with times
leadership in bennett coleman and company limited in Analytics and risk domains while
heading these domains for various organization under the group. Here I worked Followed
on I worked with Think analytics and followed up in Jupiter as head of lending and risk. Till
now I have created various products across B2B and B2C for customer/Anchor assessment,
Scorecards, cross-sell/upsell, Dynamic pricing, dynamic targets, business process
reengineering and management.

In my leasure time I love to read, write, and play TT.

I am Rahul Kumar, Vice President and Head of Risk at Arthmate.

I am having over 12 years of experience in analytics, strategy, and risk management, I have
had the privilege of working in various domains and organizations, honing my skills along the
way.

I am an alumnus of IIM Calcutta and hold a degree in computer science engineering from BIT
Sindri.

My professional journey began as the Head of Analytics for Naaptol, an exciting e-commerce
organization. From there, I transitioned to the prestigious Tata Group, where I worked as a
strategist in the telecom domain. I then had the opportunity to lead strategy, pricing and risk
domains for Bennett Coleman and Company Limited (AKA Times group). Later, I joined Think
Analytics and Jupiter as the Head of Lending and Risk contributing to their success.

Throughout my career, I have developed a diverse range of products for both B2B and B2C
segments. These include customer and Anchor assessment tools, scorecards, Alternated
data based risk assessment products, cross-sell/upsell strategies, dynamic pricing models,
dynamic targets, business process reengineering solutions and multiple risk models and
strategy.

Outside of work, I find joy in reading, writing, and indulging in a spirited game of table
tennis. These activities help me recharge and stay inspired.

Dear Jigyasa,

Hope you will be good. Today I had a word with Tanuj (CRM M3M: contact you have shared
for refund purpose). He informed me that my refund is still under process and might take a
longer time. Tanuj is the fourth person after Mahima, Bhavna and harish we talked with
(Unfortunately other three never received our call after giving commitment to update us).
Now, the problem is we have initiated refund by 28th April’23 and after multiple conversation
and documentation at 16th June M3M CRM confirm that every approval is already received
and we will get the refund by 30th June but when we enquired we got another excuse from
their side as approval form taking money to the account is pending. We have provided the
same and again we received the date of 26th July (As you communicated to us after talking
with M3M team . Now after today’s conversation I almost lost the hope and as I am under
deadly need of the money while M3M it just taking time so that they can earn more and
more interest of the hold money.
Let me remind you my background: Rahul Kumar an Alumni of IIM Calcutta, Ex edit head
Times of India, Current vice president: A large NBFC.

Now, It seems like it’s a forced measure that I have to channelize other mediums including
legal action, posting to public medium, writing article in news paper. Please note that we
had been forced to do the booking before RERA approval received to M3M and we had been
asked to share the EOI way after our 3 payment has been completed.
I believe going through this route will hamper my time as well as credential for both M3M
and your organization.
So please consider this as my last polite request and please get my refund process ASAP (as
the revised promised time was end of July) and during this month only .

Dear Jigyasa,

I trust this message finds you in good health. Today, I had a conversation with Tanuj (CRM
M3M: contact you provided for refund), who informed me that my refund is still in process
and may take longer. Tanuj is the fourth person after Mahima, Bhavna, and Harish with
whom we have spoken (unfortunately, the other three did not respond after committing to
update us).

The issue lies in the fact that we initiated the refund on 28th April’23, and after multiple
discussions and documentation, M3M CRM confirmed on 16th June that all approvals were
received, assuring us of receiving the refund by 30th June. However, upon further inquiry,
we were given another excuse, stating that approval for the transfer is still pending. We
provided the necessary documents, and once again, we were given a new date, 26th July (as
you communicated after speaking with the M3M team).

As I am currently in dire need of the money, it appears that M3M is taking an extended time,
possibly to earn more interest on the held funds.

Let me remind you of my background: I am Rahul Kumar, an Alumni of IIM Calcutta, Ex-Editor
Head at Times of India, and currently Vice President at a large NBFC.

Unfortunately, it seems that I may need to explore other avenues, including legal action,
public exposure, writing an article in the newspaper, and giving the case to recovery agency.
Please note that we had been forced to do the booking before RERA approval received to
M3M and we had been asked to share the EOI way after our 3 payment has been completed.

However, I hope to avoid such actions, as they may negatively impact both M3M and your
organization.

Hence, I humbly request your urgent attention to expedite the refund process, as the revised
promised time was by the end of July, preferably during this month itself.
Your immediate assistance in resolving this matter will be greatly appreciated, and it will
help maintain a positive relationship between our organizations.
Thank you for your understanding and cooperation.
Please write a good SOP from text given below. Please mark that other than generic
paragraph the process for Anchor evaluation and onboarding for risk sharing as well as non
FLDG model clients in tabular RACI framework with column as Activity, sub activity,
dependency, remarks.

Anchor onboarding:
 Business team will share the term sheet
Risk sharing model:
 Business/credit team will share the current rating information for anchor
 If rating for anchor will be above “AA-“ no assessment will be require and anchor will
have exposure as per the risk protection
 If
 Business team will share the Anchor’s required document with risk team
 Risk team will run Anchor assessment and share the approval with expected credit
loss on current portfolio

 Business team will share the business case with risk team including expected month
on month business, colending framework and collection responsibility
 Risk team will evaluate basis on FLDG paying capability and define the max line to be
shared with finance and Ops team
Please note that ECL written here will not the same ECL to be submitted with pricing
committee

Onboarding:

 Risk sharing model:

1. Business team will share the policy with credit and risk
2. Business team will share the product note with credit, risk and collection
which includes:
1. Customer journey
2. Data collection
3. Repayment cycle
4. Repayment method
3. Both the teams will do the assessment independently and share the
observations
4. Business team will take a go ahead with Anchor or setup the discussion if any
disconnect to resolve
5. Credit team will document the final credit policy for coding
6. Risk team will codify the BRE
7. All stakeholder monthly discussion will be done on every first week to
measure if FLDG is sufficient from loss coverage perspective based on C-
Score and actual performance
 Non FLDG model:
1. Business team will share the product note with credit, risk and collection
which includes:
1. Customer journey
2. Data collection
 3. Repayment cycle
4. Repayment method
5. Collection responsibility
2. Business team will share the pricing approval sheet with the information of:
1. Agreed line to the Anchor
2. Agreed NPA
3. Agreed cost of funds
4. Agreed sourcing fee
5. Agreed collection fee
6. Any other fee to be shared with the Anchor/sourcing agent
3. Risk team will define the solution
4. Integration team will confirm the feasibility with existing system
5. If a development is required: Risk and integration team will create a
requirement to execute the solution
6. Business team will raise the requirement to product team and get the
execution timeline
7. It will be a joint effort from Risk, Product and Teach team for system
development led by PM
8. Integration team will do the Anchor onboarding
9. Risk team will define the offer and provide offer API (Can be integrated with
selector)
Please note that no checks from colander perspective will be coded in offer
and those checks will be the part of selector API
10. Risk team will share the offer API credentials to the integration team
11. Business and Integration team will make sure that all APIs as per solution will
be integrated
12. Integration team will perform UAT testing with risk, and credit for policy
13. Anchor go live will be done
14. All stakeholder monthly discussion will be done on every first week to
measure the performance based on approval and DPD buckets
 Manual customer credit risk evaluation process:
1. Manual customer evaluation process can be done only with the approval
from business head, Credit head and risk head.
2. Manual process will always have an end date with not more than 1 month if
solution is available and not more than 2 months if development is required
3. Manual process can run in 2 ways:
1. Risk team will create a calculator and share with credit team and
further credit team will run the approval process based on calculator
output
2. Risk team will run the process by themselves
1. In this case risk team will provide manual case tracker access
to the business team
2. Business team will enter the case details once the case will be
received
3. Risk team will provide check the document sufficiency and
provide input for documents pending in the same sheet
 4. As multiple teams requests for manual cases evaluation so
cases will be evaluated based on FIFO method (only once
complete document will be received)
5. Risk team will provide ETA in the same sheet
6. After case evaluation, the output of case will be given in the
same sheet
4. On the given end date manual cases will be stopped unless and until
justifiable reasoning for the delay and new timelines will be provided.