Aayush Gupta

I’m a Data Scientist with 5.5+ years of professional experience, specializing in Plaksha University, Punjab Generative AI and Agentic AI systems for enterprise decision support. I build explainable, auditable AI that turns “black-box processes” into transparent, trustworthy ones people can actually verify and control. My focus is on human-in-the-loop agentic systems and deterministic reasoning that supports defensible decisions in complex environments.

As Lead Data Scientist at Mphasis, I led a project that converts playbooks/SOPs into structured JSON, transforms them into ontology-driven Knowledge Graphs, traverses them via Langchain using ontology-fixed logic, and tests traversal against the original SOP to flag mismatches. I designed transition-state-based traversal logic for deterministic reasoning over Knowledge Graphs, replacing ad-hoc lookups with structured navigation. I also built a Knowledge Graph-powered QA system for business users—combining relational business context queries, direct data lookups, and advanced analytics answers like ML-based forecasting, what-if simulations, and anomaly detection.

Before that, at Tibil Solutions, I built an NL2SQL chatbot that helps business users ask questions in plain English and receive SQL-backed answers, reducing time spent on routine data pulls. Earlier, as Assistant Manager at JSW Steel, I hit 94% equipment uptime by building data-driven inspection and repair strategies across a 5 MT greenfield project, and I helped commission and integrate major production equipment on time. I bring this applied, statistical rigor from my mechanical engineering background into AI systems where decisions need to be accountable—not just accurate.