Scribd
Upload
135 views16 pages

Konvergen AI - Build An AI Company

This document describes building an AI startup called Konvergen.ai. It introduces the founding team led by CEO Lintang Sutawika and lists their credentials and background in machine learning. It then discusses automated data capture solutions and an AI product lifecycle approach involving data collection, labeling, training and production. A case study on plate number optical character recognition is presented to illustrate Konvergen.ai's value proposition, system design considerations and an approach incorporating human-in-the-loop processes.

Uploaded by

hendroshadi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
135 views16 pages

Konvergen AI - Build An AI Company

This document describes building an AI startup called Konvergen.ai. It introduces the founding team led by CEO Lintang Sutawika and lists their credentials and background in machine learning. It then discusses automated data capture solutions and an AI product lifecycle approach involving data collection, labeling, training and production. A case study on plate number optical character recognition is presented to illustrate Konvergen.ai's value proposition, system design considerations and an approach incorporating human-in-the-loop processes.

Uploaded by

hendroshadi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 16

Build an AI Startup

Lintang Sutawika - Co-Founder and CEO


+6285781521554
[email protected]
linkedin.com/in/lintang-sutawika
Automate Step Zero with
Ready-to-Use APIs

Automated data capture solutions that enables less human interactions 2


Founding Team 3

TIMOTIUS DEVIN - COO LINTANG SUTAWIKA - CEO ROAN GYLBERTH - CTO


Masters Degree in Machine Learning Masters Degree in Machine Learning Masters Degree in Machine Learning
3 years research in Computer Vision Machine Learning Generalist Research in Optimization Algorithms
Background in Computer Science Background in CS and Robotics Background in CS and Informatics
Ex-Backend Engineer at Universitas Kaggle Competitions Expert Rank Certified Network Engineer
Indonesia
Trusted by Leading Businesses

Financial Services Loyalty Programs Retail

Legal

4
Talk Composition

● AI Value Proposition
● AI Life Cycle
● Case Study
AI Value Proposition

Save Money Make Money


● Boost efficiency ● Increase sales
● Better operational ● Increase engagement
accuracy
● Increase Productivity
AI Value Proposition

https://www.insurancebusinessmag.com/asia/news/breaking-news/ai-to-assist-in-determining-whos-at-fault-in-car-accidents-157846.aspx
AI Product Life Cycle

Labeling Training

Data Collection Production


Building an AI Product

Education Proof-of-Concept Production


- Explain the benefits and - Address low-hanging fruits - Take to account the life cycle of
limitation of implementing an AI - A minimum viable product that the AI system
Use Case can showcase benefits - Address how to handle corner
- Discuss the feasibility of a Use cases
Case
Case Study: Plate Number OCR

B2647TZD
Plate Number OCR, Value Proposition

- Automatically Register Car in Parking Lot


● Increase productivity,
● Lower operational cost

- Verify Insurance Registration


● Faster processing
● Increase Customer Satisfaction
Plate Number OCR, System Design

Plate Detection

Crop OCR

B2647TZQ
Plate Number OCR, Design Question

Different Plate Color? Angle Limitation?


Plate Number OCR, System Design
Plate Detection + Type Classification

Crop
OCR

B2647TZQ

Private Vehicle
License Plate?
Plate Number OCR, Considerations
- Human-in-the-loop
● 100% Accuracy is Impossible
● Incorporate people in the whole automation process in points
where AI can fail or is unsure
● Allows to periodically create new annotated data

- AI Failpoint Identification
● Can use uncertainty sampling
● Add different machine learning models
● Manual rules can help!

- Developing AI Systems
● Breakdown system to the typical AI tasks, i.e classification,
detection etc.
● When should the system be retrained?
Thank You
Lintang Sutawika - Co-Founder and CEO
+6285781521554
[email protected]
linkedin.com/in/lintang-sutawika

Source: https://xkcd.com/1838/

You might also like