BS in Statistics and Data Science

2022 – 2026

Indian Institute of Technology Kanpur

CPI: 9.0/10

Awards: Academic Excellence Award (2022 & 2023)

Honors Track Student

Minor: Machine Learning and Applications

Grade XII

2022

Delhi Public School, Surat

Score: 97.6%

Grade X

2020

Delhi Public School, Surat

Score: 98.4%

Nalanda Merit Scholarship 2020

Fee waiver worth INR 3,00,000 for securing 98.4% marks in AISSE

BYJU's Merit Scholarship 2020

Fee waiver worth INR 1,50,000 for excellent academic performance

Barker's DP-SGD

July'25 - Present

Prof. Dootika Vats, IIT Kanpur

• Developing a differentially private SGD variant using a Barker's proposal-inspired robust gradient scaling
• Improving utility-privacy tradeoffs in model training, achieving faster convergence on similar privacy guarantees

Coordinate Descent for LAD Estimation

Aug'25 - Present

Prof. Debasis Kundu, IIT Kanpur

• Developing a novel coordinate-wise descent strategy to compute LAD estimates for linear regression parameters
• Demonstrating stability in high-dimensional regimes (p > n) and performing on-par with simplex-based solvers

Languages: R, C, C++, Python, LaTeX, MATLAB, HTML, JavaScript, CSS, SQL

Technologies: Bloomberg Terminal, Fusion360, Gazebo, SKLearn, Matplotlib, Quarto, RShiny, PyTorch, TensorFlow

* Outstanding Performance | ** Excellent Performance

Machine Learning & Algorithms: Data Structure and Algorithms, Fundamentals of Computing ★★, Introduction to Machine Learning ★, Probabilistic Machine Learning, Techniques in AI & Data Mining ★, Markov Chain Monte Carlo ★

Applied Statistics: Data Science Labs ★, Computational Statistics ★, Time Series Analysis, Linear Regression and Anova, Multivariate Analysis, Bayesian Analysis

Theoretical Courses: Linear Estimations and Modeling ★, Applied Stochastic Processes ★, Theory of Statistics ★, Real Analysis, Complex Variables, Language and Society ★★

Coordinator, Debating Society

Apr'24 - Apr'25

Media and Culture Council, IIT Kanpur

• Led a 3-tier team of 40+ students; training for competitive national and international debate tournaments
• Convened the first edition of IITKAPD with 300+ participants from 6 countries making INR 50,000 in profit

Editor, Vox Populi

Apr'24 - Apr'25

Writing and Investigative Journalism

• Led a 3-tier team of 40+ Core Group Members & 20+ Asst. Editors working on reports, infographics & videos
• Conducted IIT Kanpur's first journalism conclave, Fleet Street, with a footfall of over 1,000 participants

Associate Intern

May'25 – Jul'25

Boston Consulting Group, Mumbai - Gandhinagar

Received a pre-placement offer (PPO) to join as a full-time associate in the Mumbai office

• Partnered with Gujarat Administrative Reforms Commission as knowledge partners for policy modernisation
• Benchmarked global best practices (Estonia, Singapore, UK) to design strategic digital governance interventions
• Analyzed utilization across 1400+ PHCs to optimize healthcare service coverage and streamline delivery statewide
• Re-imagined 5+ best-in-class workflows to enable data-driven governance, improve accountability and impact
• Implemented e-governance policy, impacting 60M+ citizens through enhanced digital access and service efficiency

Research Intern

May'24 – Jul'24

IIM Bangalore, Prof. Sharkarsan Basu

Got research featured in the Research Symposium on Finance and Economics 2024 by IFMR

• Optimized lending strategies & establish equity trends for public and private sector banks in economic crises
• Analyzed public and private lending data for over 100,000 firms and top 50 banks to train a predictive model
• Validated the state double-engine government hypothesis on economic growth through discriminant analysis
• Analyzed dividend stickiness by analyzing agency cost issues with large shareholders and corporate governance
• Analyzed executive structures of 17,000 firms to assess their impact on company growth & dividend stickiness

SWAP Regression

Aug'23 – Sep'25

Prof. Sharmishtha Mitra, IIT Kanpur

Advancing towards publication in collaboration with Prof. Mosuk Chow, Dept. of Statistics, Penn State University

• Developed an EM based Label-imputation mechanism to predict response – predictor roles for SWAP regression
• Implemented a weighted LAD M-step (L₁ loss) with MAD-based scale updates for a stable fit robust to outliers
• Validated the approach on USD/INR – SENSEX data outperforming standard model with 81.6% reduction in RMSE
• Auto-imputed alternating causality regimes that align with established empirical findings and economic theory

PHASR

Sep'23 – Apr'24

Prof. Indranil Saha, ERA, IIT Kanpur

First Indian team to qualify for the RoboCup MSL Challenge out of 100+ international applicants

• Designed & developed robots capable of autonomously playing football using real-time vision and swarm robotics
• Worked in a 2-tier team with 13 members to implement electronic and mechanical design for the swarm of robots
• Developed subsystems such as solenoid based kicking mechanism and automated ball handling mechanism
• Indigenously designed and manufactured the chassis for a prototype robot capable of agile movements at 4m/s

Tempered Hamiltonian Monte Carlo (THMC)

Jan'25 - May'25

Prof. Dootika Vats, IIT Kanpur

• Designed a novel variant of HMC to enhance sampling efficiency in complex multi-modal distributions
• Incorporated adaptive tempering into leapfrog integrators, improving mode traversal over high energy barriers
• Proved theoretical guarantees of reversibility and volume preservation under the proposed THMC dynamics
• Achieved superior sampling coverage across 20-mode targets and Neal's Funnel, outperforming standard HMC

The Knight and Bishop Algorithm

Jan'25 - May'25

Research Project

• Developed a hyperparameter tuning scheme for Magnetic HMC using dual averaging and recursive exploration
• Proved the invariance and ergodicity of the Magnetic HMC kernel, necessary for a valid MCMC sampler
• Identified critical gradient-based failure modes by benchmarking the algorithm on complex multi-modal targets

No U-Turn Sampler

Jul'24 - Nov'24

Prof. Dootika Vats, IIT Kanpur

• Built a solid foundation in advanced MCMC and implemented the No-U-Turn Sampler (NUTS) from first principles
• Mastered the recursive algorithm for adaptive path-length construction for hyper-parameter tuning
• Validated the sampler by replicating experiments on complex targets, confirming its efficiency over HMC