Laya Das

PhD-trained AI specialist and Technical Lead with 10+ years of experience building and shipping high-performance machine learning, deep learning, and agentic AI systems from research to production.

Specialized in designing robust multi-agent LLM architectures, scalable RAG pipelines, and reliable industrial AI solutions that handle high concurrency with low latency.

Core Technical Focus & Expertise
* Enterprise AI & Agentic Systems: Multi-agent orchestration, LLM workflows, RAG, hallucination and reliability evaluation, guardrails, and async high-concurrency backends.
* Industrial & Applied AI: Computer vision, deep learning, graph neural networks (GNN), uncertainty evaluation, and explainable asset monitoring.
* Engineering Stack: Python, PyTorch, PyTorch Geometric, LangGraph, LangChain, FastAPI, Redis, Docker, and GCP.

Track Record & Client Impact
* Production Multi-Agent Platform: Served as sole technical owner and AI Lead for a stealth AI venture, shipping an async multi-agent core backend to closed beta in under 6 months while cutting experimentation time by 60% through custom evaluation pipelines[cite: 2, 3].
* Grid & Infrastructure AI: Developed operational AI inspection systems for Swissgrid achieving high AUC with minimal labelled samples, alongside applied research projects for AspenTech, Kistler Instruments, and Tetra Pak[cite: 3].
* Academic & Research Rigor: Author of 20+ peer-reviewed publications with 1200+ citations and h-index 17 in machine learning, computer vision, and graph learning (Postdoc at ETH Zurich and Columbia University; PhD from IIT Gandhinagar)[cite: 2, 3, 4].

Whether you need to architect production-grade agentic systems, set up modular evaluation frameworks, or bring senior technical leadership to your AI roadmap, let's connect to discuss your project.