• Contributing to a peer-to-peer distributed inference network leveraging Dockerized nodes with Petals/Ollama, RAG workflows with Pinecone/Chroma, and Open-WebUI for interactive, scalable, and privacy-preserving LLM deployment.

• Finalzed a real-time speech recognition system leveraging Flask REST APIs for audio processing, React.js frontend for live transcription, and PostgreSQL for metadata storage, deployed through Docker, cutting latency by 60%.

• Directed AI email assistant using Next.js dashboard, Node.js backend, and Pinecone vector database; integrated semantic search across company emails using LlamaIndex, decreasing email search time by 75% for 200+ employees.

• Automated the newsletter publishing pipeline using TypeScript, React.js interface, and Node.js with Claude API integration; configured MySQL content management and AWS RDS scaling, reducing manual effort by 80%.

Spearheaded fraud detection system featuring Spring Boot backend, MongoDB storage, and REST APIs for real-time scoring, processed 50K+ daily transactions streamlining detection time by 50%.

• Engineered e-commerce platform using Django/PostgreSQL backend, React frontend, and AWS EC2/S3 deployment, deployed Redis caching and query optimization for 1000+ products, achieving 25% faster page loads.

• Collaborated with healthcare team to build patient data management system leveraging React.js frontend, Flask APIs, and PostgreSQL database, automated record validation and alert workflows, handling 1M+ records at 99.9% uptime.

• Constructed credit assessment platform built on Java Spring Boot APIs, Angular frontend, and MySQL database, created officer dashboard with risk calculation tools and deployed using Docker, enhancing review efficiency by 45%.

• Designed customer analytics dashboard featuring Express.js APIs, React.js frontend, and MongoDB storage, established WebSocket real-time updates, decreasing load time from 5s to 1.5s.

• Implemented personalization platform using Python/Flask backend, React frontend, and PostgreSQL database, configured user preference storage with database indexing on user recommendations, accelerating page load time by 60%.

• Administered property inspection app powered by Next.js interface, Node.js APIs, and MySQL database, digitized paper-based inspection checklists with photo uploads and automated reporting, optimizing documentation time by 75%.