May 2026/Fullstack/course
Chatbot FSM Pro
Most chatbot projects fail because of system complexity, not the AI model. When conversation flow, retrieval logic, frontend interaction, and backend services evolve without strict boundaries, the codebase quickly turns into an unmaintainable mess. Building Chatbot FSM Pro was not about tweaking models. It was a systems integration challenge.
To solve this, I engineered a multi-service architecture that explicitly separates concerns. I used a Finite State Machine (FSM) to strictly control conversational flow and NLU intent routing, preventing the AI from hallucinating out of bounds. The backend orchestrates these flows via FastAPI, feeding context to a RAG-based retrieval engine, while the frontend consumes a clean API.
The real impact was not just a working chatbot. It proved that strong interface orchestration across heterogeneous services is far more valuable than any standalone component. Coordinating these distinct layers taught me how to architect for scale rather than just building a toy demo.