ajk216212

I built a full-stack AI Snack Suggester that recommends snacks based on a user’s mood. The frontend collects the mood and sends it to a Node.js + Express backend, which uses an ONNX-optimized instruction-tuned LLM (Qwen2.5-0.5B-Instruct) via Hugging Face Transformers.js to generate a clean, human-like response. The model runs locally for fast and private inference without exposing internal reasoning. Through this project, I learned how to integrate LLMs into real applications, handle model compatibility and generation tuning, and design a simple API-driven AI system suitable for production use.

I’m working on my first project! This is so exciting. I can’t wait to share more updates as I build.