https://anubhavsinghgtm.com I engineer AI-powered analytics platforms and deep learning classification systems — orchestrating custom LLM flows with Gemini, FastAPI, and Astro. en-us Sat, 20 Jun 2026 07:02:02 GMT https://anubhavsinghgtm.com/blog/The-Invisible-Token-That-Break-My-AI https://anubhavsinghgtm.com/blog/The-Invisible-Token-That-Break-My-AI Sat, 13 Jun 2026 00:00:00 GMT A few weeks ago I wrote about debugging truncated responses from Gemini. I thought I fixed that bug, but a new feature exposed it again. This is what happened next. https://anubhavsinghgtm.com/blog/The-Bug-That-Taught-Me-More-About-LLMs-Than-Any-Tutorial https://anubhavsinghgtm.com/blog/The-Bug-That-Taught-Me-More-About-LLMs-Than-Any-Tutorial Sun, 07 Jun 2026 00:00:00 GMT Debugging my AI analytics assistant Fluxora. The LLM just... stopped. Here's what I learned about prompt engineering, context length, and why LLM debugging is fundamentally different from traditional software.