AI / ML Engineer · Quant Analyst

Engineering
intelligence
that ships.

I build production-grade AI systems — RAG pipelines, NLP classifiers, and full-stack ML platforms — and apply the same rigor to quantitative markets, where models meet live capital.

StatusOpen to roles ’26

FocusLLMs · RAG · Quant

NowQuant @ Futures First

CGPA8.77 / 10 — VIT

BasedIndia · Remote

78% directional accuracy · FOMC short-rate prediction $2M+ notional book contribution 92% code-similarity accuracy · CodeBERT 95% AAPL price prediction · LSTM · IEEE 65% faster retrieval · RAG pipeline 1.6M samples · sentiment classifier 1,000+ users onboarded · Hackathon Runner-Up 78% directional accuracy · FOMC short-rate prediction $2M+ notional book contribution 92% code-similarity accuracy · CodeBERT 95% AAPL price prediction · LSTM · IEEE 65% faster retrieval · RAG pipeline 1.6M samples · sentiment classifier 1,000+ users onboarded · Hackathon Runner-Up

The Brief

A final-year Computer Science engineer at VIT with a conviction that the best models aren't the ones with the highest validation accuracy — they're the ones that survive contact with reality. Production traffic, live markets, real users.

My work sits at the intersection of applied machine learning and systems that have to actually work. I've shipped RAG architectures into company workflows, trained sentiment models that move trading decisions, and published peer-reviewed research on sequence models for financial prediction.

Currently trading short-term interest rate markets as a quant — where a model being wrong has a P&L attached to it. It keeps me honest.

Languages

PythonC++Java TypeScriptJavaScriptRSQL

AI / ML

LLMsRAGTransformers BERTPyTorchTensorFlow Siamese NetsLSTMScikit-learn

Systems & Cloud

React / NextNode / ExpressFlask PostgreSQLMongoDBPinecone AWSGCP / VertexDocker

Experience

Quant Analyst Intern

Futures First — US Fixed Income & STIRs

Jan 2026 — Present

India

Currently here

Analyzed and executed spread strategies on Short-Term Interest Rate markets, interpreting CPI, NFP, and FOMC minutes to inform live trading across 5+ instruments daily.
Integrated quantitative analysis with fundamental research to find statistical arbitrage across Eurodollar, SOFR, and Fed Funds futures — contributing to strategies spanning a $2M+ notional book.
Built a Python NLP sentiment classifier labeling Fed communications as hawkish/dovish using FinBERT; backtested on 3 years of data for 78% directional accuracy on post-FOMC short-rate moves.
Developed market-microstructure models on order-flow imbalance, cutting manual monitoring by 40% and improving signal latency by 200ms.

AI / Automation Engineer Intern

Goldsoil Realty

May 2025 — Jun 2025

Texas, USA · Remote

Architected a Retrieval-Augmented Generation pipeline on n8n, indexing 50+ Google Drive documents into a Pinecone vector DB with a GPT-4 agent for semantic search — 65% faster retrieval.
Built a serverless chatbot using Google Apps Script and NotebookLM for structured SOP navigation across 20+ procedures, cutting manual search effort 50% and improving first-response resolution 3×.

Selected Work

AI Legal Research & Case Management Platform

React · Express.js · PostgreSQL · OpenAI · AWS Textract

Full-stack LegalTech platform for case-precedent research, document processing, and auto-generated legal drafts — ~60% less manual research time across pilot users.
Engineered a hybrid search engine fusing vector embeddings and BM25 via Reciprocal Rank Fusion, with LLM reranking — +35% top-3 retrieval precision over baseline.
Built a document Vault with AWS Textract metadata extraction and a context-aware RAG chat assistant supporting 10+ document types.

AI Code Similarity Checker

Python · Flask · React.js · Google Gemini · CodeBERT

Plagiarism-detection tool using a Siamese Network with CodeBERT encoders for semantic code similarity.
Achieved 92% accuracy on a 10K-pair test set across Python, Java, and C++ samples.

Full-Stack Blogging Platform

React · Vite · Node.js · Express · MongoDB

Production platform scaling to 100+ active users with a RESTful API, JWT auth, full CRUD, and a rich-text editor.
Maintained 99% uptime on Vercel with a responsive React/Vite frontend.

Research

ML Algorithms in Stock Market Prediction: A Performance Evaluation

Aditya Prashar et al. — LSTM model on Apple Inc. (AAPL) achieving 95% accuracy across 24-hour, 1-hour, and 5-minute prediction intervals.

IEEE ICDCC
2024

Sentiment Analysis of Social Media Data Using Deep Learning

Led a 4-member team; trained classifiers on 1.6M Twitter samples; achieved 82% accuracy with a hybrid Bi-LSTM + attention model.

Deep Learning
Team Lead

Education

Vellore Institute of Technology

B.Tech, Computer Science & Engineering

Sep 2022 — May 2026 · Tamil Nadu, India

8.77 / 10 CGPA

Eicher School, Parwanoo

Class XII · CBSE

2022 · Himachal Pradesh, India

86%

Honors

CodeNovact Hackathon — Runner-Up

Pocket Friendly · Personal Finance App

Onboarded 1,000+ users within 2 weeks; improved retention 35% via AI-driven expense categorization and personalized investment nudges.

Machine Learning Certification

Google for Developers · SmartBridge

Certified

Engineeringintelligencethat ships.