I'm a Generative AI Engineer with hands-on experience building agentic systems, LLM orchestration pipelines, and production-grade ML deployments. I specialize in LangGraph-based multi-agent coordination, MLOps, and deep learning — from research to deployment in high-compliance environments.
- 🏢 Currently: Generative AI Engineer @ xLM – Continuous Intelligence, automating GxP compliance workflows with LangGraph agents
- 🎓 B.Tech in Computer Engineering (AI specialization) — CGPA: 3.84 / 4.0 — NMIMS MPSTME, Mumbai
- 🏆 Top 1% (274th / 74,824) in Amazon ML Challenge 2024
- 📄 2× IEEE Published — AQI prediction from satellite imagery (InCoWoCo 2025) + Early Diabetes Prediction (APCIT 2024)
- 🌱 Currently exploring: AI agent memory architectures, LLMOps, and context engineering
- ⚡ Fun fact: When I'm not building AI systems, I love staying active and dancing 🕺
| Project | Description | Stack |
|---|---|---|
| AI Therapist | LLM-powered therapy agent with emotion control & memory — 70% session consistency improvement | LangGraph, FastAPI, Supabase, Vector DB |
| LipReader AI | Lip-to-text model with 87% character accuracy, 25% faster inference via TorchServe | PyTorch, OpenCV, MLflow, Docker |
| AQI Prediction | Dual-view attention model forecasting AQI & 6 pollutants — 93% R² accuracy, 35% cloud cost reduction | PyTorch, GCP, WandB, FastAPI |
| WS Buddy | Multimodal chatbot with voice, image & context memory — 85% task success rate | LangGraph, Whisper, Qdrant, Diffusion |
| Outreach-Ace | Gen-AI resume analyzer & cold email generator | LangChain, LLM, ChromaDB, Streamlit |
| Image-Lingo | Image captioning with attention mechanism, 91% accuracy | PyTorch, AWS, MLflow |
- The Hidden Cost of Agent Memory: What Mem0, Zep, and Letta Don’t Tell You
- Reward Design Is the Hard Part: Building Verifiable Rewards for Tool-Using Agents
- LangGraph Checkpointing Is Not Free: A Production Postmortem
- GRPO in Production: The Failure Modes Nobody Writes About
- Mechanistic Interpretability Is Having Its Moment: What Engineers Actually Need to Know
- Reasoning Models Don’t Reason the Way You Think
➡️ Read all articles on Medium
- 📄 IEEE InCoWoCo 2025 — Estimating Ground-Level Air Quality Index from Satellite Imagery
- 📄 IEEE APCIT 2024 — Examining ML Approaches for Early Diabetes Prediction
- 🥇 Top 1% — Amazon ML Challenge 2024 — Ranked 274th out of 74,824 participants
- 🎖️ 4th Place — genai.works Hackathon 2025 (July) — with the xLM team
AI / ML
AI Agents & LLMs
MLOps & Cloud
Databases
📫 yuvalmehta.728@gmail.com · 🌐 yuval-mehta.vercel.app