AI/ML Engineer with production experience across cloud-native infrastructure — Kubernetes, Terraform, CI/CD — now designing multi-agent systems with human-in-the-loop approval and evaluation harnesses.
A multi-agent system that triages Kubernetes incidents from Prometheus/ELK signals, diagnoses root cause with cited evidence, and proposes remediations gated behind human approval before any write reaches the cluster. Scored against an eval harness for diagnosis and remediation accuracy, with its own tool-call latency and success rate instrumented in Prometheus.
An adaptive tutoring agent combining Bayesian Knowledge Tracing and SM-2 spaced repetition to drive personalized skill selection, with a Socratic hint agent constrained in code to never reveal answers outright. A synthetic-student eval harness measured a 19-point true-mastery lift over a static-curriculum baseline on an identical practice budget.
A retrieval-augmented generation system for document Q&A — ingestion, chunking, embedding, and semantic search, returning answers with citations back to source. Being containerized for portable deployment.
A multi-agent system handling customer banking queries — loans, policy questions, fraud checks — with LangGraph orchestrating tool-calling agents that query a SQL database and external APIs.
An end-to-end churn pipeline: feature engineering, model training, and experiment tracking in MLflow, served through a FastAPI endpoint for real-time prediction, containerized with Docker.
Open to AI Engineer, ML Engineer, and Cloud DevOps roles. Based in Hyderabad, India — happy to work remote or relocate.