AI Infrastructure · Agent Systems · Platform
Thomas Sprayberry
I build and run an entire software studio on a self-hosted, autonomous AI agent fleet — and I ship the hard infrastructure underneath it: LLM proxying, agent security, and the reliability work that keeps autonomous systems honest.
Everything below links to a real repository or a real post. No slideware.
Selected work
What I've built, in the open.
-
askalf — a self-hosted AI workforce platform
An autonomous agent fleet that runs a whole software studio end to end: intake → tickets → specialist agents → pull requests, with its own nervous system, shared memory, and guardrails. I designed it and I operate it daily.
-
dario — your Claude subscription, as an API
An OAuth LLM proxy that replays Claude Code's wire shape, so any tool — Cursor, Cline, Aider, the Agent SDK — runs on your subscription pool instead of per-token bills. One local endpoint; your traffic stays in the pool.
-
Own Your Stack — the agent-security trilogy
Three composable controls for letting an agent touch real systems without handing it the keys. Vet the tool, contain the call, give it a credential it never holds.
- warden — a deterministic, offline action firewall that decides what a tool call may do before it runs: green/yellow/red/black risk tiers, secret-exfil and prompt-injection blocking, tamper-evident audit. /warden ↗
- canon — the supply-chain gate: vet, sign, and pin every skill and MCP server before it loads. /canon ↗
- keeper — an agent secrets vault that hands out scoped, short-lived, single-use leases instead of raw keys, and audits every access. /keeper ↗
-
deepdive — a research agent you can trust by construction
One command, one cited answer: plan → search → headless fetch → extract → synthesize, every call through your own router. Built on deterministic trust signals, because citation-verification alone scores content farms a perfect 1.00.
github.com/askalf/deepdive ↗ why citation-checking isn't enough ↗
More open source — picket (a prompt-injection firewall for any CDP browser), cordon (a PII-redacting LLM gateway), hands (a cross-platform computer-use agent), claude-sync (portable Claude Code sessions), hybrid (a local-first LLM router), amnesia (self-hosted private search). All repositories ↗
Writing — receipts, not think-pieces
Every post is a system I built and broke.
An engineering blog at sprayberrylabs.com/blog — 15+ posts, each grounded in real infrastructure rather than opinion.
About
I work at the layer where AI meets real infrastructure.
Rather than write about agents, I run a fleet of them in production — and I build the proxying, security, and reliability primitives that make that safe to do. The repositories and posts above are the work itself, not a portfolio of it.
I'm drawn to the unglamorous parts: the firewall that has to be deterministic and offline, the proxy that has to stay honest when a provider pulls a model overnight, the audit trail that has to be tamper-evident. If autonomous systems are going to touch anything that matters, that's the work that has to exist first.
Contact
Open to senior AI-infrastructure, platform, and agent-systems roles.
- GitHub
- github.com/askalf ↗
- hello@sprayberrylabs.com
- Studio
- sprayberrylabs.com ↗
- Résumé
- PDF — or by email