Arjun Gupta
AI Engineer · Agentic Systems · Knowledge Graphs · Async Data Pipelines · Voice AI
Building production agentic systems end-to-end: federated multi-tenant orchestration, hybrid retrieval over knowledge graphs, and async Python infrastructure for real-time pipelines.
About
AI Engineer focused on shipping production agentic systems end-to-end. I architect federated multi-tenant platforms with stateful LangGraph orchestrators and custom MCP servers, design async-first Python infrastructure (FastAPI, asyncpg) with event-driven message buses, and build hybrid retrieval engines combining PostgreSQL pgvector with Neo4j knowledge graphs and semantic reranking. Previously built low-latency voice AI systems with LiveKit, PipeCat, and Deepgram.
Experience
AI Engineer Intern
Oct 2025 – PresentStealth AI Startup
Sole full-stack engineer and product owner across three production codebases. Architected a federated multi-tenant agentic platform with Hub/Spoke A2A communication, stateful LangGraph orchestrators, and a custom 14-tool MCP server for dynamic tool execution. Engineered an async-first Python stack (FastAPI, asyncpg) with an event-driven message bus handling real-time WhatsApp WebSockets and Meta Cloud APIs. Built a 4-way hybrid retrieval engine combining PostgreSQL pgvector (HNSW) with a Neo4j knowledge graph, layered with HyDE and Cohere reranking and an LLM-as-judge (PageIndex tree) framework for hallucination control.
AI Engineer Intern
Mar 2025 – Jul 2025ZuduAI
Developed production-grade AI agents using LangChain, LangGraph, and RAG in a microservices setup. Designed and deployed low-latency voice AI systems with LiveKit and PipeCat, integrating Deepgram STT and multilingual TTS pipelines. Automated workflows with Apache Airflow and n8n, and built post-call NLP analysis pipelines using Azure Cognitive Services.
Skills
AI / ML
Languages
Backend / Infra
Audio / Voice
Achievements
Top 5 in Fetch-a-thon (AI-based EV Trip Planner)
Finalist in Code For Bharat Season 1 @ Microsoft
Global Rank 2567 in LeetCode Biweekly 144
Top 10 in MLSA MIET Hackathon
Selected Work
A collection of projects showcasing expertise in AI/ML, voice agents, and full-stack development.
Real-time conversational AI voice agent with RAG pipeline using LlamaIndex and LiveKit. Integrated OpenAI GPT-4o, Deepgram for STT, and ElevenLabs for multilingual TTS.
Multi-faceted AI platform providing comprehensive support to farmers. Features RAG-powered chatbot for government schemes, ML-based crop recommendation engine, and real-time voice AI agent.
Autonomous AI agent system to discover, process, and provide structured information about sports tournaments using agentic workflows and dynamic query generation.
Modern, integrated environment for coding-related activities with TypeScript frontend for responsive UX and robust Python backend for server-side logic.
Standalone, containerized microservice for intelligent chat functionality, designed for easy integration into larger systems with Docker deployment.
Comprehensive logistics platform for Indian Army fleet and armament management with real-time GPS tracking, predictive maintenance, and digital KOTE system.
Fully serverless web application that processes uploaded CSV data and visualizes results as interactive bar charts using AWS cloud services.
Get in Touch
Currently seeking full-time opportunities to leverage my skills in AI/ML engineering, conversational AI, and voice agent development. Let's connect!