_{Interactive site · Next.js · GSAP · Three.js · Framer Motion}

Enterprise GenAI Engineer with 4+ years designing and shipping production-grade LLM systems for regulated enterprise workflows. I work where retrieval quality, structured outputs, governance, and cost control matter — building systems that pass finance, procurement, and compliance review without losing engineering rigor.

Currently building production GenAI at Jio Platforms (Reliance) — LLM-assisted financial decision workflows with deterministic fallbacks, RAG over policy documents with role-based access controls, and a centralized LLM evaluation/governance framework with prompt versioning, drift detection, hallucination guards, and rollback. Pursuing M.Tech, Software Systems at BITS Pilani (WILP).

• Decision Support — LLM-powered systems with deterministic fallbacks, confidence-based routing, and human approval workflows for compliance-sensitive use cases.
• RAG & Retrieval — Retrieval-Augmented Generation pipelines over enterprise data with hybrid retrieval (BM25 + dense vectors), reranking, and strict role-based access controls.
• Structured I/O & Streaming — JSON-mode and Pydantic schemas plus SSE token streaming, so model output can flow safely into downstream backend systems and real-time approval consoles.
• Evaluation & Governance — Offline regression suites, LLM-as-a-Judge scoring, drift and hallucination detection, prompt and configuration versioning, and rollback strategies.
• Event-Driven Integration — GenAI services wired into event-driven backends with retries, circuit isolation, semantic caching, rate limiting, and trace-level observability.
• Cloud-Native Guardrails — Audit logging, PII redaction, jailbreak and topic-boundary defense, and granular cost and token-budget enforcement.

GenAI & LLM Engineering

RAG & Retrieval

LLM Ops & Governance

AI Frameworks & Agents

Backend & Systems

Data & Storage

Frontend

Infra & Engineering Practices

_{Reliability-first · Audit-first · Production-first.}