// since 1990
I design systems that
age well.
Philosophy background. 35 years building infrastructure. Distributed systems, AI orchestration, and the tooling that makes it all work.
Currently building something interesting.
// what I do
Distributed Infrastructure
High-availability clusters, replicated storage, automatic failover. The stuff that needs to work at 3am without calling anyone.
AI Systems
Multi-model orchestration, agent architectures, production deployments. Making LLMs reliable enough for real workloads.
Developer Tooling
CLIs, APIs, deployment pipelines. The invisible infrastructure that makes engineering teams move faster.
// open source
github.com/cognitioncommons →LLM developer tools. The utilities I built because I needed them and figured others might too.
token-budget
Count tokens and estimate costs across LLM providers
prompt-diff
Diff prompt templates with semantic awareness
llm-cache
Cache LLM responses for development and testing
guardrail-tester
Test LLM outputs against safety guardrails
model-compare
Compare outputs across different LLM models
+ 11 more
Response validators, retry clients, cost trackers...
// the long arc
First code. BASIC on a 286. Before the web existed.
Slackware on floppy disks. Got X11 working when that was an achievement.
Philosophy degree. Epistemology, philosophy of mind, logic.
Enterprise infrastructure. Cloud migrations. Distributed systems at scale.
Building something at the intersection of all of it.
// research
Cognition Commons
Open research on cognitive architectures and multi-agent coordination. Documenting patterns for building AI systems that reason.
cognitioncommons.org →Research areas:
- Multi-agent coordination
- Tool-using agent architectures
- Alignment techniques
- Uncertainty in AI systems
// stack
Languages
- Python
- TypeScript
- Go
- Rust
Infrastructure
- Docker / Kubernetes
- Terraform
- Linux (since '93)
- HA clustering
Data
- PostgreSQL
- Redis
- Elasticsearch
- Vector databases
AI/ML
- LLM orchestration
- Agent frameworks
- RAG systems
- Fine-tuning
Available for hard problems.
Consulting engagements for infrastructure, AI systems, and developer tooling. Short-term, high-impact work. No full-time roles.