█████╗ ███╗   ███╗██████╗ ██╗████████╗
██╔══██╗████╗ ████║██╔══██╗██║╚══██╔══╝
███████║██╔████╔██║██████╔╝██║   ██║   
██╔══██║██║╚██╔╝██║██╔══██╗██║   ██║   
██║  ██║██║ ╚═╝ ██║██║  ██║██║   ██║   
╚═╝  ╚═╝╚═╝     ╚═╝╚═╝  ╚═╝╚═╝   ╚═╝  

I work at the intersection of LLM research and production ML — evaluating language models at scale, building reasoning systems, and shipping data products that reach millions of users.

• 🔭  Researching multilingual LLM performance gaps and cross-lingual reasoning at ASU
• 🏦  ex-Wells Fargo · Built GenAI QA systems and LLM evaluation frameworks on the Consumer GenAI team as Quant Analytics intern
• 🏢  ex-Bajaj Finance · Shipped churn prediction, notification recommendation engine, and forecasting pipelines to production
• 🎓  B.Tech CS @ VIT Vellore · Merit Scholarship, Top 10 rank
• 🏆  Runner-up @ Voxel51 Visual AI Hackathon  ·  Finalist @ JPMorgan Code For Good

Identifying LLM Performance Gaps Across Languages
Benchmarked OLMo-2-7B across ~100 languages and 10 datasets. Measured the correlation between pre-training token exposure and downstream accuracy, and tested English Chain-of-Thought as a cross-lingual reasoning bridge.

OLMo-2  vLLM  NLLB-200  Hunyuan-MT  FastText  Self-Consistency

Self-Taught Reasoner for Mathematical Problem Solving
Implemented the STaR methodology on LLaMA 3.2-3B — iterative rationale refinement with hint-guided correction, outperforming both vanilla SFT and zero-shot CoT on GSM8K.

LLaMA-3.2  vLLM  TRL  SFT  Chain-of-Thought

Autonomous Grocery Shopping Agent
Conversational AI agent that converts natural language into checkout-ready grocery carts via Instacart API, Knowledge Graphs, and LLM orchestration with automated checkout using BrowserUse and Playwright.

LLM Agents  Knowledge Graphs  BrowserUse  Playwright

AI / LLMs

Languages & Data

MLOps & Tools