Research Scientist at Meta Reality Labs specializing in machine learning acceleration and SIMD vectorization. https://mohankku.github.io/ Sun, 06 Sep 2026 00:51:59 +0000 Sun, 06 Sep 2026 00:51:59 +0000 Jekyll A chat page on a static GitHub Pages site that talks to a local Ollama model sounds contradictory: there is no server to hold API keys, run tool loops, or keep secrets. I built exactly that for this site’s /chat page, and the constraints shaped every decision. Here is what worked, what bit, and the architecture that fell out. Sat, 05 Sep 2026 00:00:00 +0000 https://mohankku.github.io/2026/09/05/local-first-chat-static-site.html https://mohankku.github.io/2026/09/05/local-first-chat-static-site.html agentslocal-aiollamatool-callingprivacyjekyll Autonomous coding agents have rapidly evolved from simple chat completions into multi-turn loop runners that read, edit, execute, and verify software repositories. While cloud-hosted frontier models (Claude 3.7 Sonnet, GPT-4o) excel at architectural synthesis, relying entirely on remote APIs presents distinct tradeoffs: strict telemetry/privacy concerns for proprietary codebases, cumulative API costs, network latency, and zero offline availability. Fri, 04 Sep 2026 00:00:00 +0000 https://mohankku.github.io/2026/09/04/zenith-local-coding-agent.html https://mohankku.github.io/2026/09/04/zenith-local-coding-agent.html agentslocal-aisystemsollamacoding-assistantmemory Getting a neural network off a workstation GPU and onto a phone or headset NPU is not one step — it is a pipeline of graph transformations, each with its own optimizations. This post walks the full path: an FP32 PyTorch model, through quantization and operator fusion, down through lowering to a delegated NPU binary with ExecuTorch. Fri, 04 Sep 2026 00:00:00 +0000 https://mohankku.github.io/2026/09/04/from-fp32-to-npu.html https://mohankku.github.io/2026/09/04/from-fp32-to-npu.html quantizationexecutorchpytorchNPUedge-ML