Architected and deployed a production-grade GenAI chat system using AWS Bedrock (Claude), PostgreSQL, and a RAG pipeline, serving 50K+ real-time queries and driving 40% higher user engagement.

Built an LLM-powered allocation engine using ChromaDB and LangChain Agents, introducing a smart semantic search layer for workforce-project matching that reduced manual effort by 60%.

Built end-to-end MLOps pipelines with automated model validation, versioning, and performance monitoring to ensure stable production inference.

Developed and deployed a predictive analytics-based Mental Health Toolkit using Nest.js, MongoDB, Azure, and Redis, enabling 25% faster clinical decision-making and 15% improvement in revenue cycle efficiency; recognized with SPOT Monetary Award.

Built and tested scalable backend services for ACH transaction processing for a fintech client using Java, Spring Boot, and AWS DynamoDB, integrating internal APIs for real-time verification and simulating 10K+ concurrent users; reduced API errors by 30%.

Implemented secure authentication (OAuth2, OIDC, JWT, SAML) while owning CI/CD pipelines on Azure with Docker and Kubernetes, ensuring reliable, production-ready deployments.

Processed high-volume streaming sensor and video data, building distributed pipelines and deploying containerized services on AWS using Docker and Kubernetes for low-latency analytics.

Designed and trained machine learning and computer vision models, performing data cleaning, feature engineering, and model evaluation to achieve 90% accurate live crowd density estimation.

Led a team of 10 engineers to build an intelligent virtual office collaboration platform, integrating ML-driven user behavior analytics and predictive models to forecast resource utilization and team interactions, reducing physical office expenses by 35%.

Created a Visitor Management System to streamline in-store check-ins and monitor customer visits.