Computer Engineering @ K.J. Somaiya College of Engineering
Software Engineering • AI/ML • Agentic Systems • Open Source
I'm a Computer Engineering undergraduate at K.J. Somaiya College of Engineering, interested in building reliable software systems at the intersection of software engineering, AI/ML, backend systems, and developer tooling.
- 🔭 Currently building agentic AI and retrieval systems
- 💼 Machine Learning Intern at Larsen & Toubro
- 🌱 Exploring RAG, multi-agent systems, AI evaluation, scalable backend systems, and developer infrastructure
- 🛠️ Open-source contributor to Microsoft APM, TypeAgent, and Google Benchmark
- 🏆 Hackathon winner/finalist with experience taking ideas from prototype to working systems
- 📍 Mumbai, India
Jun 2026 – Present
- Built an end-to-end invoice intelligence pipeline using Python and PaddleOCR PP-StructureV3 for extracting metadata and line-item tables from multi-page documents.
- Engineered vendor-agnostic extraction with confidence calibration, schema validation, cross-field consistency checks, structural repair, and exception handling.
- Implemented AI governance controls including confidence-based guardrails, human-in-the-loop review, PII controls, audit logging, and drift monitoring.
May 2025 – Jul 2025
- Developed and deployed 10+ production-grade React pages, reducing user drop-off by 30%.
- Engineered a backend RAG-based document processing pipeline, improving retrieval relevance by 40%.
May 2024 – Sep 2024
- Built a MERN research submission platform, reducing review time by 40%.
- Optimized APIs and database schemas, improving query performance by 35%.
- Refactored a monolithic CLI into a modular package architecture.
- Fixed command regressions and improved CLI consistency while maintaining stability across 6,700+ tests.
- Built a pipeline-aligned embedding benchmarking system using MRR and Hit Rate.
- Implemented min-score sweeps and fixed correctness issues in semantic-search evaluation.
- Added opt-in cross-thread statistics for multithreaded counters.
- Implemented mean, median, standard deviation, and coefficient-of-variation statistics for analyzing thread-level performance imbalance.
Agent-Based Market Intelligence & Trading Simulation
Designed a 6-agent AI swarm combining market data, financial news, and macroeconomic context for trading analysis.
OpenAI Tavily Redis Socket.io React Node.js
- Built a low-latency market-data pipeline with sub-10 second updates.
- Designed specialized agents for collaborative financial analysis and decision support.
Podcast Knowledge Retrieval
Built and benchmarked retrieval pipelines for querying long-form podcast transcripts in TypeAgent.
Python Embeddings Semantic Search MRR Hit Rate
- Benchmarked multiple embedding models.
- Implemented semantic retrieval, ranking, and systematic retrieval evaluation.
AI Financial Advisory Platform — Mumbai Hacks Finalist
Agent-based financial advisory platform for investments, tax analysis, and cash-flow insights.
AI Agents Pinecone Vector Search RAG
- Built an agent-based financial analysis workflow.
- Improved retrieval speed by 50% using Pinecone vector search.
RAG Systems • Agentic AI • Semantic Search • Embedding Evaluation • Scikit-learn • TensorFlow • NumPy • Pandas
- 🥈 2nd Place — Revengers Hack 2026
- 🚀 6th Place — Monad Blitz Mumbai 2026
- 🏆 Winner — IETE Student Exploration Challenge 2026
- 🤖 Finalist — Gemini HackDays
- 💻 Rank 6 — MLSA Competitive Programming
I'm interested in software engineering, AI/ML engineering, agentic systems, developer infrastructure, and open-source collaboration.