About
I build backend systems that make AI products work in production: RAG pipelines, AI agents, LLM integrations, and the APIs that connect them. At Alteam I shipped a semantic search feature with LangChain and PGVector that cut talent search time by ~40% and saved long-term costs by ~35%. My core stack is TypeScript, Node.js, Python, and FastAPI, with hands-on experience in RAG systems, agent orchestration, and vector databases - plus fullstack capability in React and Next.js.
What I offer
I offer specialized AI engineering services to help you build and scale your AI applications.
RAG & Agent Development
Design and build retrieval pipelines using PGVector and LangChain, including multi-index search, tool use, and agent routing. Connected to your Postgres database with proper access control and audit logging.
AI Feature Integration
Full-stack implementation of AI features into your product using React/Next.js on the frontend and Node/Express on the backend. Deployed on cloud platforms like Vercel and AWS, with CI/CD pipelines setup.
AI Observability Setup
Instrumentation for cost tracking, latency monitoring, and prompt/config CI/CD so your AI system doesn't become a black box in production.
Security & Permissions for AI
Ensuring your AI only accesses data a user is authorised to see. Authentication, row-level access, structured logging, and role-based access.
Professional Work Experience
Skills
Core skills
Deep Dives
Detailed explorations of complex problems and my architectural solutions.
Check out my latest work
I build AI systems that actually ship. I specialize in developing RAG AI agents. My projects span full-stack AI deployment: from vector database and agent orchestration to API integration and production monitoring. Below are systems I have built end-to-end.
Talent Sourcing Agent (RAG)
A single-agent RAG system that helps clients find the best talent from Alteam's database for their needs by using AI to understand their requirements and match them with the best candidates.
AI News Aggregator (GenAI)
An AI-driven news aggregation service that curates, summarizes and emails a daily digest of the latest AI news.
Scalable Real-Time Chat App
A scalable real-time chat system with a Node.js backend, using Redis Pub/Sub for real-time messaging, Kafka for message processing, PostgreSQL for durable storage, and NextAuth.js for secure authentication.
StudyLM - Create Study Notes with AI
An AI-powered note-taking application built with Next.js where users can create, edit, and organize notes, backed by a Convex data layer. It integrates LangChain and Gemini to generate summaries and insights from notes, while Clerk provides secure authentication and Tailwind CSS delivers a modern, responsive UI.
App Review Classification Using PLMs
An end-to-end NLP system that classifies app reviews using fine-tuned models like BERT4RE, ALBERT, and DistilBERT, achieving over 89% F1-score. A Flask-based backend and UI let users submit single or batched reviews for classification and export results to Excel, helping product teams analyze feedback efficiently.
Teaching Tech
During my time in VJTI, I was a part of two communities that were mainly focused on learning new skills and teaching others the same.
- G
Google Developer Student Clubs VJTI
Mumbai, India
Organized and led Android app development sessions, teaching over 10 students Kotlin, Android Studio, and CRUD app creation. Conducted workshops on unit testing, app architecture, and best practices, while guiding students in using Git and GitHub for effective version control and collaboration. - C
Community of Coders VJTI
Mumbai
Mentored 8 junior developers across two teams, leading a full-stack web app project using the MERN stack and guiding a cross-platform mobile app development with Flutter and Firebase. Conducted code reviews, resolved issues, ensured successful deployments, and organized knowledge-sharing sessions on web development, app development, and machine learning.
I love participating in Hacks
During my time in VJTI, I attended 3+ hackathons. Students from Mumbai would come together and build incredible things in 2-3 days. It was fun to build and collaborate with motivated and passionate individuals.
- G
General Mills CodeIcon
Mumbai, India
Led a team in a 2-day ML hackathon, analyzing a recipes dataset to predict cuisine types using Naive Bayes and Random Forest models. - A
ACM Hackthon
Mumbai, India
Collaborated with a team to develop an HR Management System aimed at streamlining administrative workflows such as employee data management, leave requests, and reimbursement claims. - H
Hack With Code Warriors
Online
Developed Boctor, a trained conversational chatbot with images(link) support which does not allow you to feel absence of doctors and nutritionists and provides important health related information. It suggests diet plans based on user's age, weight, height, gender and the level of pysical activeness(or the no. of days a user works out). By just having a conversation with the bot, the users can book appointments with doctors. It has other features like displaying 3D images of heart, pills, etc.