Hi, I'm Rafid Tahmid.
I turn real problems into working software.

Type any topic and get an interactive diagram: an ordered learning path from foundational to advanced, where clicking a node generates its explanation on demand, with per-node chat grounded in open sources.

• Two-model split — a strong reasoning model plans the learning-path structure while a cheap RAG-grounded model writes node descriptions on demand; structures cached per topic
• Multi-source RAG — pgvector similarity search before generation; on cache miss, just-in-time fetch/chunk/embed from six sources (Wikipedia, Wikibooks, arXiv, PubMed, Stack Exchange, MDN) with inline numbered citations
• Score-based model routing — similarity ≥ 0.72 routes to the cheap model, below escalates to the stronger model; low-confidence answers auto-retry on a stronger model
• Provider-agnostic orchestration across Anthropic, OpenAI, and Google behind one interface with automatic promotion and retry on failures
• Live cost dashboard — token spend by model, cache savings, and % served from DB vs freshly generated

Connect Gmail and an AI agent (Claude Opus as the brain, plus tools) scans the mailbox, identifies subscription receipts, and extracts recurring spend into a clean dashboard — monthly spend charts and per-subscription cards with total spent, next payment, and overdue alerts.

• Privacy-first — read-only Gmail access; stores parsed facts only, never raw email bodies. Messages are read in memory during a scan and discarded
• Agent pipeline — a cheap heuristic search narrows the mailbox to billing-signal candidates before the agent classifies and extracts
• CI pipeline with GitHub Actions

Built collaboratively with Shafayet

Meal planning and recipe management web app: discover recipes, plan weekly meals, and generate shopping lists with real-time price comparisons across major Australian supermarkets. University group project — scored 96%.

• Backend developer role — designed the MySQL schema and built a secure backend with Node.js, Express.js, and bcrypt authentication
• API integration with TheMealDB, Coles, and Woolworths for real-time price comparison and dynamic shopping lists
• Responsive design with dark mode support

Academic research papers written for the Grand Challenges in Computer Science course — research and writing samples showing analytical breadth alongside engineering work.

• “Blockchain for Secure Online Transaction”
• “Stock price reactions to dividend announcements from major mining companies (BHP, Rio Tinto) on the ASX”