Building systems where agents plan, act, and reason.
Recent B.E. Computer Science graduate specializing in LLM-powered multi-agent systems, RAG pipelines, and agentic AI — from research prototype to deployed cloud service.
Turning agentic AI research into working products.
I specialize in designing multi-agent systems using frameworks like LangChain, LangGraph, CrewAI, and AutoGen — orchestrating agents that plan, execute, and review each other's work rather than relying on a single monolithic prompt.
My work spans the full stack: retrieval-augmented generation pipelines backed by vector databases (ChromaDB, FAISS, Pinecone, Qdrant), model deployment on AWS (SageMaker, Lambda, API Gateway), and shipping the result as an actual usable interface with FastAPI, Streamlit, or Flask.
Most recently, I interned at Botree Software building ANN-based forecasting models on real production and sales data — bridging classical ML with the agentic systems I build on the side.
The stack behind the agents.
Languages
Agentic / LLM Frameworks
RAG & Vector DBs
ML / DL
Cloud (AWS)
Deployment & Tools
Selected work.
ATS Resume Checker
Code →- Multi-agent ATS analyzer built with CrewAI and Groq LLM that parses resumes (PDF, DOCX, HTML) and scrapes job description URLs for automated resume-JD alignment.
- Generates a multidimensional match report using keyword extraction, cosine similarity, and semantic similarity scoring, with actionable improvement suggestions.
VacAIgent — Agentic Travel Planner
Code →- Multi-agent task-planning system using CrewAI that generates structured, day-wise travel itineraries based on user goals and time constraints via intelligent agent decomposition.
- Served through a FastAPI backend with an interactive Streamlit dashboard for context-aware, LLM-driven scheduling.
Analyzer GPT — Multi-Agent Code Analyst
Code →- Dual-agent AutoGen system: a Code Executor runs submitted code securely inside a Docker sandbox, then hands the output to an Analyzer agent.
- The Analyzer agent reviews the result using RAG-based, context-aware Q&A, and routes it back to fix errors if any occur — surfaced through an interactive Streamlit interface.
Movie Recommendation Engine
Code →- Content-based filtering engine using cosine similarity, with compressed pickle + gzip storage for efficient similarity-matrix handling at scale.
- Integrated the TMDb API for real-time poster fetching; deployed as a Flask web app on Render.
Where I've worked.
AI/ML Engineering Intern
Botree Software Pvt. Ltd. — Gurugram, India- Conducted exploratory data analysis on enterprise production and sales datasets, uncovering trends that directly informed quarterly strategic planning.
- Designed and trained an ANN model to forecast production and sales performance, improving prediction reliability over baseline heuristics.
- Delivered data-driven insights to senior stakeholders, influencing supply-chain and inventory decisions.
Academic background.
B.E. Computer Science
Bharati Vidyapeeth College of Engineering — Pune, MaharashtraClass XII (CBSE)
M.R.L Sr. Sec School — DelhiLet's build something.
Open to full-time AI/ML Engineer roles and freelance work involving LLM systems, RAG pipelines, or agentic automation. Reach out directly — I usually reply within a day.
Say Hello →