AI Engineer | Problem Solver | Solution Builder

Building AI that actually does something useful.

I'm a final-year CS undergrad who got obsessed with LLMs, agents, and everything in between. I like taking messy real-world problems and turning them into systems that work fast, reliably, and at scale.

Right now I'm most excited about Agentic AI, RAG pipelines, and voice-first applications. If it involves making an AI system think rather than just respond, I'm probably already reading papers about it at 1 AM.

• Agentic AI & Automation Workflows : Building multi-agent pipelines (LangGraph, CrewAI) where agents plan, execute, and self-correct across complex tasks. If a human is doing something repetitive, I'm probably already thinking about how to automate it.

• Generative AI Systems (Text → Anything) : Turning a raw idea into images, audio, or video - no blank canvas, no creative block. End-to-end generative pipelines that produce consistent, controllable, production-ready creative assets.

• Full-Stack AI Integration : Building the complete stack around AI, async task queues (Celery + Redis), cloud-deployed backends (FastAPI on GCP), vector and relational databases. Currently doing this end-to-end in my photo retrieval system.

• RAG & Knowledge Systems : Hybrid retrieval (dense + sparse), query rewriting, metadata filtering, embedding caching. Shipped RAG systems that went from 14s latency to 2s. I treat retrieval as an engineering problem, not a prompt engineering one.

• Data-Driven Problem Solving : I work backwards from a metric. Whether it's a 0.90 silhouette score or a 7x latency reduction, I use ML to solve a specific, measurable problem.

🎥 VartaVerse - AI Automation System for Media Enterprises (Under-Development 🚧)

Building an end-to-end multi-agent AI pipeline that converts a trending news topic into scenes as well as short explainer video.

LangGraph · FastAPI · GCP · Gemini LLM · Streamlit

🎙️ Arthsakshar AI — Multilingual Voice Agent for Budget Literacy

An AI voice platform (STT/TTS) that autonomously handles outreach calls in multiple languages. Built for a government initiative targeting 65,000+ Chartered Accountants across Maharashtra. Reduced estimated human agent workload by ~80% for routine outreach.

Sarvam AI · Twilio · Gemini · PostgreSQL

🗃️ QueryMyPDF — Hybrid RAG Chatbot

A multi-turn PDF Q&A system with a FAISS + BM25 hybrid retrieval setup 2x better retrieval precision than single-retriever systems, 0% failure rate on 1000+ page documents, sub-second latency.

LangGraph · FAISS · BM25 · Gemini 2.5 Flash · Streamlit

🤳 LookItUp - Media Management and Retrieval System

An end-to-end face-recognition photo retrieval system using InsightFace (ArcFace) and FAISS vector search, enabling event attendees to find their photos from 500+ image albums via a single selfie. Designed a privacy-first architecture storing only 512-dim face embeddings (no raw face crops), with async batch processing via Celery + Redis.

Python · Redis · Celery · Arcface · Supabase

🛍️ Customer Segmentation & Recommendation Engine

Personalized recommendation system deployed as a production-ready FastAPI backend for real-time, dynamic ad-tech requests.

K-Means · PCA · FastAPI · EDA

Lead AI Developer @ Stealth AI Startup (Oct 2025 - Present)

AI Engineer @ Kartavya Technology, Bengaluru (Jan 2026 – Present)

Software Developer @ Scoutripper (Jul 2025 – Dec 2026)

Languages       Python · C · C++ · SQL
Gen AI          LLMs · RAG · Agentic Workflows · Embeddings · Vector Databases · Fine-tuning LLMs(LoRA/PEFT)
Frameworks      LangChain · LangGraph · CrewAI · HuggingFace · OpenAI API · OpenCV 
ML/DL           PyTorch · TensorFlow · Scikit-learn · CNN/RNN/LSTM · Model Optimization
Deployment      FastAPI · Flask · Docker · Google Cloud · Streamlit
Databases       PostgreSQL · Supabase · Pinecone · Neo4j · MySQL

• B.Tech (CSE) — IKG Punjab Technical University, Jalandhar (2026)
• XII — Pradhan Public School, Agra (81%)

• 🥇 Top 10 — redBus Data Decode Hackathon
• 📜 Develop Gen AI Applications — IBM
• 📜 Advanced RAG with Vector Databases — IBM
• 📜 Machine Learning — NPTEL & IIT Madras
• 📜 Supervised Machine Learning — DeepLearning.AI & Coursera
• 📜 Claude Certifications — Anthropic
• 📜 Business Intelligence — Infosys

• LinkedIn: linkedin.com/in/chaudharypwn/
• Email: chaudharypawan1779@gmail.com

I'm always open to discuss project ideas, collaborations, or just a good conversation about AI.

"I don't just build demos. I build things that work."