Jay Javeri

Quantitative Researcher

About

Hi, I’m Jay. I build machine learning systems that sit close to research but are designed to work in real-world settings. Recently, I built a multimodal vision language system combining world model embeddings with LLMs for spatial reasoning, training on 400K samples and running large-scale ablations across multiple benchmarks. I’ve also worked on financial ML problems using 800K+ data points and 500+ earnings call transcripts, improving predictive performance by 10 to 15 percent through feature engineering and model design. In another project, I built an agent-based evacuation simulator modeling hundreds of agents under real-world constraints, achieving up to 33 percent improvements over baseline strategies. Beyond projects, I’m a Senior Teaching Assistant for Georgia Tech’s algorithms course, supporting 1600+ students on topics like dynamic programming, graph algorithms, and correctness. I’m most interested in problems where: models need to work outside ideal conditions evaluation is unclear or imperfect strong empirical and statistical reasoning matters I’m currently exploring roles as a Machine Learning Engineer, Research Engineer, or Member of Technical Staff across startups, frontier labs, and quantitative teams. If you’re building something at the intersection of ML systems, research, and real-world deployment, I’d love to connect.

Country

City

United States

Industry

Computer Software

Skill

Machine Learning, Artificial Intelligence (AI), Full-Stack Development, Engineering Data Management, Statistical Data Analysis, Research, Programming, Java, C++, Python (Programming Language)

Experience

Quantitative Researcher

Arctic Light Capital

2024-5 - 2024-12 · 8 mos

Mumbai, Maharashtra, India

Formulated and empirically evaluated hypotheses on correlation- and cointegration-based trading signals using 9 months of OHLC data across NIFTY 50 equities. Applied rolling-window statistical analysis and ADF-based cointegration tests to assess signal stability under realistic market regimes. Identified low persistence and non-robustness of classical statistical signals, motivating exploration of option-implied volatility models (e.g., Black–Scholes) and alternative alpha sources.

Software Engineer Intern

Javeri Fiscal Services Ltd

2022-5 - 2024-8 · 2 yrs 4 mos

Mumbai, Maharashtra, India

Built Python-based automation tools for end-of-day holdings reporting and multi-trade price aggregation, reducing processing time by 80%. Improved reliability and turnaround time of downstream financial analysis through automation and structured data processing.

Research Assistant

Georgia Tech Financial Services Innovation Lab

2023-5 - 2023-7 · 3 mos

Atlanta, Georgia, United States

Designed reproducible data pipelines for Compustat and CRSP, supporting CAPM beta estimation and Fama–French 48 industry classification across 50+ years of financial data (1.6M+ cells). Developed and evaluated supervised machine learning models for corporate credit rating prediction using financial ratios and SIC-based industry features. Conducted applied financial ML research under the mentorship of Dr. Agoston Reguly, emphasizing empirical evaluation and data quality.

Education

Georgia Institute of Technology

Computer Science

2025-4 - 2026-5 · 1 yr 2 mos

Pursuing a Master of Science in Machine Learning CS 8803 - Efficient Machine Learning CS 8803 - Vision Language Models CS 7641 - Machine Learning CS 6515 - Graduate Algorithms CS 7643 - Deep Learning CS 7637 - Knowledge based AI