Soumajit Das

About Soumajit Das

Bridging the gap between Industrial Systems and Generative AI.With over 6 years of engineering experience in large-scale industrial projects (ITER Nuclear Fusion) and advanced research in Industrial AI at IIT Madras, I specialize in building robust AI systems. I combine the rigor of industrial system design with the flexibility of modern Generative AI.My expertise lies in transitioning research prototypes into production-grade solutions, specifically within Root Cause Analysis (RCA), Multi-Agent Systems, and Data Engineering.

Experience

1. Artificial Intelligence Engineer

   Tata Consultancy Services

   Jun 2019 — Present · Pune, IN

   Core Technical Focus: Generative AI & Agents: Architecting Neuro-Symbolic frameworks that combine Deep Learning (Transformers) with LLM Agents (Grok, Llama 3) for reasoning and validation. Experienced in Multi-Agent RAG and MCP (Multi-Channel Processing) concepts. Industrial Data Engineering: Building petabyte-scale data pipelines using PySpark and SQL, ensuring 100% data integrity for critical infrastructure. Edge AI & TinyML: Deploying optimized object detection models (FOMO) on constrained edge devices (Nicla Vision) with <100ms latency.Current Research:Developing a Human-in-the-Loop Diagnostic Assistant for HVAC systems. This framework utilizes Temporal Causal Discovery (TCDF) and LLM-based Semantic Validation to solve \"Symptom Explosion\" in sensor data, achieving 97.9% diagnostic accuracy on the LBNL benchmark.I am passionate about solving \"Closed-Loop Confounding\" in control systems and building AI that doesn\'t just predict, but explains and reasons.

Education

• National Gems Higher Secondary School

  Secondary

  2010 — 2013

• Indian Institute of Technology, Madras

  Master of Technology - MTech, Industrial Artificial Intelligence

• Narayana Institute

  Higher Secondary, Engineering Science

  2013 — 2015

• University of Engineering & Management (UEM)

  Bachelor of Technology, Mechanical Engineering