cloudwright
AI-powered cloud architecture - describe infrastructure in natural language, get Terraform, cost estimates, and compliance reports
Python 32 2 7 weeks ago
Xavier Puspus
AI Architecture and Engineering
I work on the layer between language models and production: retrieval, evaluation, cloud infrastructure described in natural language. Mostly Python and TypeScript, mostly things that have to actually ship.
Pinned work
All projects-
-
kb-arena
Benchmark 19 retrieval architectures (vector, contextual, QnA, knowledge graph, LightRAG, hybrid, RAPTOR, PageIndex, BM25, metadata-filtered, temporal, rerank, HyDE, multi-query, late-interaction, SPLADE, agentic, plus two quantum-inspired rerankers) on your own docs. Citable evidence bundles, bootstrap CIs, significance tests.
Python 12 3 3 weeks ago -
ai-workflow-benchmark
Benchmark harness measuring AI coding tool+workflow performance, not just model capability. 100 tasks, sigmoid scoring, 12 capability dimensions, gap analysis.
Python 14 5 4 weeks ago -
ph-civic-data-mcp
Philippine civic data as MCP tools: the full PSA OpenSTAT statistical catalog, hazards, weather, procurement, elections, and one-call area profiles. 41 tools, no API keys.
Python 5 2 6 weeks ago -
airgap
Air-gapped AI customer support for enterprise. On-device LLM, offline-first, config-driven. 7 industry templates.
TypeScript 10 1 2 days ago -
agentforeman
Approve, steer, and stop every Claude Code agent from one local page.
Python 5 0 3 days ago
Stack
- Languages
- Python TypeScript
- AI/ML
- LangGraph MCP OpenAI SDK Anthropic SDK
- Backend
- FastAPI Postgres
- Frontend
- Astro Next.js React Tailwind
- Infra
- AWS Netlify Docker
Writing
All posts-
memory-arena: 20 agent-memory strategies, one eval
Twenty memory strategies through one lifecycle, one judge, one model. A 30-line vector store outranks every funded vendor SDK.
-
leaves-ph: measuring Metro Manila's tree canopy from orbit
A human-calibrated Sentinel-2 canopy model for 17 NCR LGUs, 2019 to 2026, scored F1 0.78 against hand-labeled pixels.
-
sinkmap-ph: how fast Philippine cities are sinking
Land subsidence for seven cities from Sentinel-1 InSAR, 2016 to 2025, mapped next to where the floods actually hit.