SONU KUMAR

I’m a data analytics-focused AI engineer and M.Tech student in Information Technology (Data Analytics). In my role at Tata Consultancy Services, I built an AI-powered Natural Language to SQL agent using the AGNO Agentic Framework, generating and executing SQL on AWS PostgreSQL with strong evaluation results (highest accuracy with minimal hallucination).

I improved performance through schema caching and conversational memory to cut latency and enable context-aware chats. I also delivered real-time systems using FastAPI, React, and Socket.IO, plus data visualization and an interactive Metabase analytics dashboard for continuous business monitoring—used in General Electric Healthcare enterprise analytics for non-technical teams.

I back my engineering with hands-on research: from EEG-based emotion recognition using SOGNN variants and fairness-oriented benchmarking, to multi-agent ad content generation, relationship chatbots with Gemini + Pinecone, recipe summarization with fine-tuned BART, and real-time AI search chatbots combining Tavily and LLMs. I value rigorous evaluation, practical deployment readiness, and building tools that help people make better decisions.