In Progress
Local RAG with Ollama
A local-first RAG pattern for private document Q&A.
Ollama, embeddings, vector DB
GitHub soonArticle
Labs
Experiments, demos, and technical proofs.
Labs captures the path from architecture notes to working prototypes, open examples, and future productized kits.
Coding Agent v0.1
Basic local coding workflow.
Fix a small Python bug using Ollama and qwen2.5-coder:7b. Preview a diff, explicitly approve the correction, and run calculator tests in a Docker container with networking disabled.
Available prototype ? Python ? Ollama ? Docker
Follow the complete setup and exercise, including agent safety tests, an expected calculator failure, review and approval, and a passing test run. The agent and Ollama run on the host; this prototype does not establish a complete air-gapped security boundary.
Runnable examples
Inspect the controls in action.
Two synthetic labs demonstrate permission-aware retrieval and release evidence. Explore their architecture, recorded outputs, tests, and downloadable source.
Experiments
Current lab pipeline
Planned
Azure OpenAI RAG
A cloud RAG pattern for governed enterprise document intelligence.
Azure OpenAI, storage, search
GitHub soonArticle
In Progress
Hallucination detection demo
Answer support checks using retrieved evidence and LLM-based revalidation.
RAG, validators, evals
GitHub soonArticle
Planned
Token monitoring dashboard
Usage, cost, latency, and evaluation telemetry for AI systems.
OpenTelemetry, logs, charts
GitHub soonArticle
Planned
Multi-agent workflow demo
A controlled agent workflow connected to approved tools and policies.
Agents, tools, APIs
GitHub soonArticle
Planned
OpenShift AI deployment experiment
Deployment manifests and operational patterns for private AI services.
OpenShift, routes, PVCs
GitHub soonArticle