Work
Real systems, described as engineering — not marketing. Three are open source under the Innovation Ways GitHub org; the fourth was a client engagement, shown here at the architecture level.
IW AI Core
Open sourceAn AI-orchestration platform that runs AI-assisted development across projects — with all operational state in Postgres and a dual-agent runtime.
- Python 3.12
- FastAPI
- PostgreSQL
- llama-index
- OSS · MIT
IW RAG
Open sourceA fully-local, single-GPU agentic RAG system with a benchmarking studio built in — grounded, cited answers over your own documents, and the numbers to prove how well any configuration actually performs.
- Python 3.12
- LangGraph
- vLLM
- pgvector + BM25
- Local · single-GPU
- OSS · MIT
PodForger
Open sourceA full-stack platform that turns RSS feeds into multi-speaker AI audio podcasts — LLM scripting and self-hosted GPU text-to-speech, running in production.
- FastAPI
- Next.js 16
- async SQLAlchemy
- Multi-LLM
- GPU TTS
- OSS
CORE — on-prem multi-agent RAG
Client engagementA 100%-local, five-agent retrieval platform built for a Swiss telecom operator: document RAG with citations and RBAC, NL→SQL behind a validation layer, code analysis, and an orchestrator.
- LangGraph
- pgvector
- Ollama
- RBAC
- On-prem
Building something that has to be right?
Selectively available for engagements where owning the architecture of a production AI or BSS system is the point.
Start a conversation