Amine Benkirane

AI-first movie and TV show recommendation platform powered by IMDB metadata and LLMs. • Designed and built the full backend: API routes, services, database, and auth (JWT) • Created personalized suggestion logic based on seen, liked, ignored titles • Built an AI assistant that understands intent, finds similar content, and filters via natural language • Integrated OpenAI for smart recommendations and conversational flow • Implemented robust ops: logging, monitoring, tests, CI/CD with GitHub Actions • Dockerized and deployed to GCP (Cloud Run) — scalable and production-ready 🔧 Stack: FastAPI · PostgreSQL · OpenAI API · Docker · GCP · GitHub Actions · JWT Auth

Conversational flight assistant capable of finding the best flight combinations within flexible date windows, tailored to user preferences. • Extracts and validates travel details using a LLM, with strong control-flow and flag management for smooth conversation handling • Custom scraping engine built on top of Google Flights: supports flexible search, optimized round trips, scoring logic, and clean API exposure • Full backend orchestration: user input → LLM parsing → flight API engine → GPT-generated response • User session handling, JWT-based authentication, and preference storage built with FastAPI • Dockerized architecture deployed to AWS (ECS + S3 + Route 53) with CI/CD pipeline ready via GitHub Actions • Basic observability built in: structured logging, error tracking, and deployment visibility 🔧 Stack: React · FastAPI · OpenAI · Docker · AWS · GitHub Actions · Google Flights Scraper

Developed a legal chatbot capable of answering procedural and legal questions using a vector-based retrieval pipeline. • Designed a complete backend architecture: session handling, embedding logic, vector search, LLM orchestration • Scraped and processed legal PDF documents into structured text, chunked and embedded via OpenAI + ChromaDB • Orchestrated the full pipeline: authentication, chat history, vector queries, prompt injection, and clean API responses via FastAPI • Dockerized the app for portability • Deployed on AWS, with CI-ready infrastructure 🔧 Stack: FastAPI · OpenAI · ChromaDB · JSON · Docker · Azure · HTML Scraping

Built a bilingual chatbot (English/Spanish) for a small personal loan company to answer customer questions about financial products and loan conditions. • Developed a RAG pipeline: vectorization, contextual search, and response generation with LLM • Structured the FAQ knowledge base into chunks, embedded using OpenAI + ChromaDB, with strict retrieval logic - Built the backend in FastAPI, including session handling, context injection, and similarity scoring • Designed a modular, fast-iteration-ready API layer • Dockerized and deployed on AWS and GCP 🔧 Stack: FastAPI · OpenAI API · ChromaDB · Docker · AWS · GCP