I'm an AI Software Engineer, GenAI Developer, Traveler and Gamer.
Expert in RAG and building chatbots. To know more about me checkout VAi.
- Featured on CNBC News for building VAi, an AI chatbot that talks to recruiters - Read the article
- 🤖 AI Software Engineer at the Division of IT, University of Maryland (Oct 2025 – Present)
- Building an end-to-end Config Evaluation feature to score chatbot configurations on demand via RAGAS.
- Reduced a manual evaluation workflow to a single click that returns 5 quality metrics in ~2 minutes, accelerating config iteration for the team.
- Refactored the DocReview platform, reducing code duplication by 70% with 90%+ test coverage
- Deployed a scalable chatbot inference service connecting GraphQL APIs, PostgreSQL/pgvector, and Neo4j knowledge graphs to improve RAG response speed and lower compute cost
- Built serverless data-sync pipelines pulling from Congress.gov and Federal Register APIs into DynamoDB
- Lead QA testing and browser automation with Playwright and Claude Code
- 🎓 Pursuing a Master of Engineering in Software Engineering at UMD (Jan 2025 – Dec 2026, GPA 3.78/4.0)
- 🏦 Previously Software Engineer I at Bank of New York and Technology Consultant Intern at PwC
- 20+ AI Software Applications Built, Deployed and Scaled
- 3+ Years of Experience Working with Fortune 500's like Bank of New York (BNY) and PriceWaterhouseCoopers (PwC)
- 18+ Degrees & Certifications That define my learnings and hands-on experience
- 10+ Blogs and Articles Sharing Knowledge to the community
- SIM Scholar - Schindler Igniting Minds Scholarship (Top 30 across India) - Sponsored my Bachelor of Technology degree at Veermata Jijabai Technological Institute, Mumbai · Oct 2019
- Publicis Sapient Jumpstart 2022 - Selected through a nationwide hackathon (Top 200 of 28,000+), Delhi · Jul 2022
- Bertelsmann Technology Scholarship - Issued by Udacity · Dec 2021
- Phish Detection ML - Multimodal phishing detection using dual-tower CNNs in PyTorch, achieving 98.99% validation accuracy across text and metadata features.
- DocuMind - Document Q&A platform built with LangChain and RAG, reducing irrelevant LLM outputs by 35% via optimized semantic search and custom embeddings.
- Codistant - Edge-native AI programming assistant on Cloudflare using Llama 3.3 70B and TypeScript, with real-time SSE streaming and persistent multi-turn histories via Durable Objects and SQLite.
- Languages
- AI & Data Science
- Databases
- Frameworks
- Tools & Platforms
