Blog
Blog
Lessons from building end-to-end AI — voice assistants, multi-agent LLMs and on-premise systems in production.
Reading a model's architecture from its safetensors header, without downloading the weights
How I recover a neural network's anatomy from tensor names and shapes alone: 96.8% role accuracy on 510 architectures, from about 46 KB of header. The method behind AI Model Explorer.
Why emotion models fall apart outside the lab (and what actually helps)
Speech emotion recognition looks solved on benchmarks and breaks on real calls. A researcher's tour of domain shift, speaker leakage, metric learning and self-supervised features.
FinOps-Estimator: sizing GPUs for a voice AI pipeline
An open-source tool that turns concurrency and model choices for STT → LLM → TTS into VRAM, RAM, CPU, TFLOPS and cloud or on-prem cost.
Fitting an LLM in your own server room: the memory math behind on-prem AI
Bits per weight, quantization error, outliers, KV cache and memory bandwidth: how to decide which model actually runs on the GPUs you own.
TurboQuant, reviewed: when a theory paper walks straight into production
A close read of TurboQuant (ICLR 2026): random rotations, Lloyd–Max codebooks and a 1-bit QJL residual, and why it went from proof to vLLM and a 17k-star vector index within weeks.
Your MAU is not your load: capacity math for voice AI at scale
How to turn "a million users" into a latency budget, a concurrency number, KV-cache memory and a GPU count — with the formulas, not vibes.