I'm an AI systems engineer focused on shipping real AI systems end-to-end — from sub-400ms real-time voice agents and hybrid RAG pipelines with Reciprocal Rank Fusion, to Model Context Protocol (MCP) servers and multi-modal placement engines.

I have a strong bias for turning prototype failure modes into measurable engineering improvements.

• 📍 Based in Bhopal, India
• 🏆 3× National Hackathon Winner (NASA Space Apps, Qubitx 2025, KV Sangathan)
• 🎓 B.Tech in CSE @ Oriental Institute of Science and Technology (2023-2027) | President of Student Activity Council
• 💼 Freelance AI Automation Engineer & Former Tech Intern @ HaronIndia Pvt. Ltd.
• 📫 How to reach me: connect.harshawardhan@gmail.com

Real-Time Multimodal Voice AI Agent for Indian Banking

• Engineered an ultra-low latency multimodal voice AI agent combining streaming STT/TTS; reduced latency from 1,250ms to ~390ms.
• Built a hybrid RAG engine combining BM25 and vector search with Reciprocal Rank Fusion (RRF); improved Hit@1 to 93.8%.
• Designed a 7-layer barge-in system with <280ms interrupt-to-silence latency.
• Tech: Python, FastAPI, WebSockets, Groq, Deepgram, Qdrant, PyTorch, AudioWorklet

MCP Server & Agentic Code Review Tool

• Built a Model Context Protocol (MCP) server exposing an analyze_repo tool over JSON-RPC/stdio.
• Designed a dual-model analysis pipeline using Gemini 2.5 Flash and Groq LLaMA 3 for fast structured scoring across 8 categories.
• Engineered asynchronous, rate-limit-aware GitHub ingestion with Docker containerization.
• Tech: MCP SDK v2, Gemini, LLaMA 3, GitHub REST API, Docker, Pydantic

Multi-Modal AI Interview & Placement Readiness Platform

• Architected an end-to-end placement readiness engine combining ATS resume scanning, GitHub codebase analysis, and computer-vision proctored mock interviews.
• Designed a 6-factor deterministic + neural ATS engine using semantic cosine similarity.
• Implemented real-time MediaPipe FaceMesh for attention tracking paired with Whisper STT.
• Tech: FastAPI, React 18, Firebase, ChromaDB, Groq, Gemini, MediaPipe, Whisper

• AI & ML: LLMs, Agentic Pipelines, RAG, Hybrid Search (BM25 + RRF), MCP, PyTorch, Whisper, HuBERT, MediaPipe
• Backend: Python, AsyncIO, FastAPI, WebSockets, WebRTC, Pydantic, HTTPX
• Infrastructure: Qdrant, ChromaDB, Firebase, Docker, GitHub Actions, Linux