Client needed a web app that makes complex medical reports understandable for everyday users — no medical background required. I built a full-stack AI-powered platform with a split-screen interface: users upload a medical report (PDF or image), which appears on the left panel, while an AI chat assistant on the right answers their questions in simple, plain language. Each session is tied to one report for accurate, focused responses. Key Features: ➤ Split-screen UI: report viewer + AI chat side by side ➤ Conversational AI that explains lab results, diagnoses & medical terms in plain English ➤ Chat history sidebar for revisiting past report discussions ➤ Secure login/signup with session management ➤ Pricing page, settings, and a clean landing page ➤ Per-report chat scoping for accurate context handling Tech Stack: ➤ Frontend: Next.js + TailwindCSS ➤ Backend: FastAPI (Python) ➤ Database: PostgreSQL ➤ AI/LLM: OpenAI GPT-4o + LangChain ➤ File Storage: AWS S3 ➤ Auth: JWT-based authentication ➤ Deployment: AWS + Vercel Delivered a fully functional MVP on time. The client's goal was to reduce patient anxiety around medical reports by making AI accessible to non-technical users — this product directly addressed that gap.

0

48

Multi-Restaurant Food Ordering Platform • Built a scalable food ordering platform (MERN) with real-time order tracking, an advanced Admin Dashboard, Role-Based Authentication, and secure Stripe-powered checkout. • Designed a responsive UI with seamless navigation, a customer reviews module, and optimized backend APIs for reliable order management.

2

92

Home Inspection Quote Platform Built a full-stack home inspection platform connecting customers with an admin for quote management. Customers submit detailed inspection request forms; admin receives and manages all quotes from a secure dashboard. Features: multi-field form with validation, role-based access control (customer vs admin), responsive design, and a conversion-focused landing page. Stack: Next.js, FastAPI, PostgreSQL, JWT auth.

0

39

Persistent RAG Chatbot: Dual-Model AI Architecture Production-grade AI chatbot using LangGraph with dual-model architecture — Groq (Llama 3.3) for speed, Gemini Flash-Lite for complex reasoning. Automated fallback logic switches models on API quota exhaustion, maintaining 100% uptime. RAG pipeline built with FAISS and HuggingFace MiniLM embeddings for offline semantic search. Chat memory persists across sessions via MongoDB. Real-time token streaming via SSE with React memoization. Next.js 15 frontend, FastAPI backend, Node.js proxy for stream handling.

0

59

Bilal Ahmad

Full Stack Ai Engineer | Ai Agents | Next.js