S HARVANI J ADHAV the-7g-girl ‡

Senior Undergraduate Sharvani Jadhav

Mathematics and Scientific Computing +91-9552365775 Ó
Indian Institute of Technology Kanpur [email protected]
Year Degree/Certificate Institute CPI/%
2021 - Present Graduation Indian Institute of Technology Kanpur 7.3/10
2021 Class XII (HSC) Abhinav College, Pune 90.3%
2019 Class X (ICSE) St. Helena’s School, Pune 95.6%
S CHOLASTIC A CHIEVEMENTS T ECHNICAL P ROJECTS
• Qualified for Indian National Maths Olympiad (INMO) 3 times, Nightlight Analysis of Factories ‡ Jun’23-Present
ranking among top 20 in Maharashtra & top 900 students in India Mentor: Prof. Srinivasan Rangan, Finance & Accounting, IIM Bangalore
• All India Rank 2488, JEE Advanced 2021 (top 1% of 2.5L+ pupils) • Analysed NASA’s VIIRS Nightlight dataset at 500m2 resolution
• All India Rank 3945, JEE Mains 2021 (top 0.4% of 1M+ aspirants) and a 24-hour frequency, extracted using Google Earth Engine
• Leveraged the EPA dataset to identify over 8,000 factories across
I NTERNATIONAL R ESEARCH E XPERIENCE the USA, and analyze the total waste disposed by them annually
Bayesian Modelling of Road Traffic Collisions ‡ Jun’24-Present • Integrated the two results via time series analysis and correlation
Mentor: Prof. Lee Fawcett, Newcastle University, England techniques like Pearson correlation and Chatterjee correlation
Received Letter of Recommendation for exceptional problem-solving skills Adversarial Attacks in Face Biometrics ‡ Jun’23-Dec’23
• Developed Accident Predictive Model using Bayesian statistics Mentor: Prof. Soumya Dutta, CSE, IIT Kanpur
to find best government strategy for reducing road accidents • Analyzed CNN architectures focusing on the impact of
• Applied AIC-based model selection to determine 7 parameters adversarial attacks on different layers of neural networks
for negative binomial regression on data from 67 traffic sites • Employed the FGSM to generate adversarial attacks that
• Implemented the Metropolis-Hastings MCMC algorithm for significantly reduced MTCNN’s face detection accuracy by 48%
10,000 iterations to precisely fine-tune model hyperparameters • Improved attack’s effectiveness by using Momentum Iterative
• Evaluated various priors (Weibull, Lognormal, etc) with different FGSM, achieving robust misclassification with 56% confidence
parameters using DIC to identify the best fit for causality data Introduction to Machine Learning ‡ Aug’23-Nov’23
• Analyzed data from 56 sites, identifying 11 high-risk locations CS771 Course Project under Prof. Piyush Rai
for targeted speed camera placement reducing casualties by 40% • Implemented kernelized ridge regression, achieving accuracy of
96.7% with an RBF kernel & optimally located landmarks
W ORK E XPERIENCE
• Applied PCA and t-SNE to project 784-D MNIST digit data to 2-D,
Generative AI Intern | Quantstac Mar’24-Jun’24 visualizing and comparing the two clustering methodologies
• Developed an AI-based video advertisement generator to • Improved classification accuracy from 46.9% to 73.3%, by using
incorporate branded items into existing movie scenes using VAEs a hyper-parameterized multi-output linear regression model
• Tuned stable video diffusion model on 10+ hyperparameters:
Image Processing ‡ Aug’23-Nov’23
sampler, scheduler, seed, cfg-min for text-to-video generation EE604 Course Project under Prof. Tushar Sandhan
• Upscaled low-resolution images to high resolution using models • Developed a cross bilateral filter algorithm using NumPy &
like SUPIR, LDSR etc in ComfyUI achieving 4.5x improvement OpenCV to balance brightness & saturation in day & night images
Machine Learning Intern | Nitrosk Jan’24-Mar’24 • Developed a de-skewing algorithm using Sobel filter & contours
• Developed a deep learning-based model to estimate body to align skewed Sanskrit sentences, improving OCR accuracy
measurements from video input for an easy contactless solution • Built an audio classification system to distinguish between
• Implemented face and body detection using OpenCV with metal & cardboard sounds using windowed Fourier transforms
background removal from videos containing up to 20+ people Business Analytics using Python ‡ May’22-Aug’22
• Incorporated gyroscopic angles to adjust the model for tilt and Finance and Analytics Club, IIT Kanpur
modeled body in 3D from measurements using GANs • Preprocessed Mergers & Acquisitions data and utilized
Investment Banking Analyst | Finlatics Nov’23-Dec’23 Matplotlib & Seaborn to generate bar, line and scatter plots
• Set up a simulated private equity fund & assessed investment • Utilised multilinear regression, decision trees and CNNs to
proposals to identify startups with significant growth potential perform financial modeling and forecast stock market trends
• Explored various investment strategies, including deal sourcing, • Developed a model for accurately predicting house prices in
valuation, negotiation, and exit strategy to maximize returns California using SVM, Random Forest Regressor & XGBoost
• Utilized the discounted cash flow model to accurately determine Game Theory and Auctions Jun’23-Dec’23
the valuation of 7 companies & to conduct in-depth case studies Stamatics Society, IIT Kanpur
P OSITIONS OF R ESPONSIBILITY • Utilized Nash Equilibrium & Multi-Strategy Nash Equilibrium to
develop optimized solutions that maximize payoffs in games
President | Hall of Residence VI, IIT Kanpur Mar’23-May’24 • Investigated strategic decision-making in game theory models,
• Elected by 1000+ residents to preside over the Hall Executive like the Prisoner’s Dilemma, Stag Hunt, and Battle of the Sexes
Committee of 7 members, supervising 110+ contract workers • Applied First-price and Second-price auctions theory along with
• Led Shauryas to secure the first position in Galaxy, an inter-hall Revenue Equivalence Theorem to solve the preset assignments
cultural competition, marking the highest rank in its history
• Installed 12 washing machines worth Rs. 4L for improves living T ECHNICAL S KILLS
• Organized Hall Day with Rs. 5 lakh attracting 2,000+ attendees • Programming Languages: C++, C, Python, SQL, R
Leader | Bicycling Society, IIT Kanpur May’22-May’23 • Utilities: MATLAB, R studio, Tableau, Excel, LATEX
• Led a team of 12 secretaries in organizing regular cycling rides • Libraries: Numpy, Pandas, Matplotlib, Scikit-Learn, Seaborn
resulting in 3x growth of cycling community post-covid • Web: HTML, CSS, JavaScript, ReactJs, Figma
• Organized first-of-its-kind Duathlon on Independence day
attracting 100+ participants, garnering 20k+ Instagram views
R ELEVANT C OURSES
• Achieved a 5x increase in online reach through social media • ML: Advanced topics in ML, Intro to ML, Image Processing
Batch Senator | Students’ Senate, IIT Kanpur Feb’22-Feb’23 • Probablity: Probability and Statistics, Bayesian Statistics
• Elected by a batch of 1200+ Y21 students and became student • CS: Data Structures & Algorithms, Fundamentals of Computing
body representative to address their queries in students’ senate • Maths: Linear Algebra, Numerical Analysis & Computing