This repository serves as a centralized hub for the things I build - ranging from fully offline AI coding agents and multi-agent frameworks, to scalable full-stack applications. Feel free to explore the code, fork the experiments, or reach out to talk tech.
The posts live at sectumpsempra.github.io.
Compiling llama.cpp with Metal GPU acceleration on an M1 Mac, loading Google's Gemma-4 26B via Unsloth's quantization, and wiring it to OpenCode for a fully agentic, offline coding workflow. Total API cost: $0. Data sent to the cloud: 0 bytes.
Four specialized AI agents collaborating in a cyclic LangGraph state machine to architect, write Terraform, audit security with remediation loops, and render architecture diagrams - from a single plain-English description.
Written by Amit Bhatt