• Built an academic workspace using NextJS for the frontend (Vercel), and FastAPI for the backend (Render). Used Supabase with PGVector for embeddings, managed using Drizzle. • Created 3 AI Agents for course planning, exam generation and concept podcast creation using Autogen and Pydantic, and deployed them on Modal for web inference. • Served the interactions using OpenRouter and Langchain. Utilised Langfuse to monitor outputs and cost.

• Implemented ETL pipelines using AWS S3/Glue/Spark and integrated with Apache Airflow, refining terabytes of transaction data for machine learning. • Deployed a custom Grafana plugin (React, Docker) enabling granular MySQL metrics filtering, accelerating dashboard query analysis by 2500% for 25 developers. • Built high-performance dashboards with NextJS on AWS Amplify, streaming from AWS RDS to handle thousands of concurrent requests efficiently.

• Developed a Workday X2 extension (React, TypeScript) with a proxy user switch, saving 7,500 hours ($500,000) annually. • Linked OMS metadata and JIRA tickets in MySQL, and used LLaMa to identify duplicate JIRA tickets. • Reduced page load times by 90% (5 min to 30s) for high-usage business assets by re-architecting with Xpresso and reorganizing key data views.