AI Engineer · Full Stack Architect · Founder @ NorahLabs
I build production AI systems — agentic orchestration engines, RAG pipelines, and native desktop applications. Based in Nairobi, Kenya. Currently focused on making serious AI tooling accessible to developers who don't want cloud lock-in or privacy trade-offs.
AWS Certified Data Engineer · Google Cloud Professional Data Engineer · AWS Certified Developer
A native desktop AI coding environment for developers who want real architectural understanding — not just autocomplete.
v8.2.0 · Mac + Linux · atlarix.dev
Atlarix parses your codebase into a live Blueprint graph via Round-Trip Engineering using oxc-parser (Rust, ~2× faster than SWC) for TypeScript/JS and WASM tree-sitter for Python, Go, and Rust — then uses that graph as RAG for every AI query, reducing token usage from ~100K to ~5K per query.
v8.2.x highlights:
oxc-parserfor TS/JS/JSX/TSX — Rust-based, 256KB per-file cap, cooperative yield every 10 filesWASM tree-sitter— Python, Go, Rust viaweb-tree-sitter+tree-sitter-wasms, packaged viaextraResources- Parser worker pool — 2–4 dedicated workers with timeout and crash recovery
get_repo_overview— lightweight cold-session orientation without full Blueprint parse- Secret path denylist —
.env*,*.pem,*.key, service account files blocked from all read paths
v8.1.x highlights:
- Blueprint restored lazy — graph schema, IPC, click-to-load canvas (500-node cap)
- Tool consolidation 60+ → 28 — unified
search(auto/text/semantic/structural),edit_file, on-demand MCP (list_available_mcps+mcp_call), Skills via ToolRegistry compress_context— model-triggered, not pre-dispatch; removes latency from every standard turnget_workspace_overview+read_attachment_summarymacro tools- Streaming guarantee — every exit path emits
chat:response:endexactly once - ~50% token reduction per turn — 6 pre-dispatch injections converted to tool-driven retrieval (verified via OpenRouter logs)
v7.x foundations:
- Parallel agents: Research · Architect · Builder · Reviewer · Debugger · Explore
- DelegationMonitor — cyclic and runaway delegation detection with bus abort
- Compass managed tier — no API key required (Fast / Balanced / Thinking)
- Full CI/CD with Apple Notarisation + Linux packaging (.deb / .rpm / .AppImage)
- Three-tier subscription (Free / $0 · Pro / $19mo · Workforce / $79mo)
- ResearchLedger-driven deep research with BM25 relevance filtering
Multi-provider: OpenAI · Anthropic · Google Gemini · Groq · Mistral · xAI · OpenRouter · Together AI · Ollama · LM Studio
Achievements: Winner — Amazon Nova AI Hackathon (Devpost × AWS) · Red Bull Basement Global Innovation Programme · GDG Nairobi Agentathon (May 2026)
atlarix-skills — Community skill registry for Atlarix. Skills teach agents language patterns, framework conventions, and developer workflows across React, Next.js, Python, TypeScript, Go, Rust, Docker, MCP, testing, and more. Apache 2.0.
atlarix-mcps — MCP marketplace index for Atlarix. One-click server connections from the in-app marketplace.
atlarix-releases — Desktop app release builds for Mac and Linux.
| Layer | Technologies |
|---|---|
| Languages | TypeScript · Python · JavaScript · SQL · Bash |
| AI / ML | Agentic Systems · RAG · Graph RAG · Multi-Agent Orchestration · Context Compression · LangChain · Hugging Face · oxc-parser · Tree-sitter (WASM) |
| Desktop | Electron · React · React Flow · SQLite |
| Backend | Node.js · Django DRF · FastAPI · PostgreSQL · Redis · Celery |
| Infrastructure | AWS · GCP · Docker · GitHub Actions · Supabase · Vercel |
| Auth & Billing | Auth0 · Token Vault · LemonSqueezy |
| Monitoring | Sentry · PostHog · Prometheus · Grafana |
Praxia — AI healthcare assistant with MONAI X-ray analysis (pneumonia, fractures, tumours), multilingual symptom diagnosis, real-time WebSocket chat, and Docker/Nginx/Celery production stack.
Commit Checker — Open source CLI tool for Git commit analysis.
Atlarix is built in Nairobi. Most serious AI developer tooling assumes a Western cloud subscription. Atlarix is open-model by design — bring any provider, including locally-built African models. Currently piloting with African AI labs for native MansaLLM integration.


