Production AI Engagements
I'm Damian Galarza. I work with engineering teams on retainer to ship reliable AI features, harden the AI stack, and make agentic systems that survive past the demo.
15+ years building production software. Former CTO at Buoy Software where we shipped FDA-cleared medical device software. Currently senior engineer at August Health building August Intelligence. The patterns I bring are tested with real users, real token budgets, and real error rates — not adapted from blog posts.
15+
years building production software
0 → 50+
engineers scaled as CTO
FDA
cleared medical device software shipped
1+ yr
daily Claude Code user, documented publicly
$12,000 · 2 weeks
A focused 2-week engagement that produces a written architectural assessment and one shipped foundation piece. Either side can walk away after, or convert to retainer.
$15k–$60k/month
Ongoing engagement at the intensity you need — Advisory, Embedded, or Full. Flat monthly rate, flexible hours, direction reset every two weeks. Two-month minimum after Foundation Sprint, then 30-day notice.
Backed by a 30-day money-back guarantee. Also available: 1:1 coaching for engineers shipping production agents and the leaders defending the spec.
"He quickly understood where we were with AI tooling and gave us immediately actionable advice, not generic frameworks. He identified gaps we hadn't considered, walked us through how he architects agent loops in production, and helped us think through our product-level agent strategy without over-engineering it."
"I found his videos especially clear-headed. I booked a couple of private sessions to discuss OpenClaw, and to my delight, he was equally clear-headed in person."
Production AI Engineering
Architecture, evals, observability, governance, and the production tradeoffs most teams skip until it hurts.
Claude Code
Context windows, MCP servers, CLAUDE.md systems, rollout patterns, and real-world development workflows.
AI Agent Architecture
Memory systems, orchestration, tool use, governance, and the production patterns behind agentic systems.
Most AI demos hand one agent the whole job. When the output touches reputation, money, or outbound email, use a workflow instead of an agent loop. Here's where the boundary goes, built with Mastra.
When an agent fails, harness engineering gives you four levers (Context, Tools, Loop, Governance) to find which one broke in under a minute.
Prompt-based approval gates fail because the model decides whether to ask. Mastra's requireApproval primitive removes that decision entirely. Here's how to implement it.
What 'supported by a fleet of agents' means in practice: which tasks automate, which don't, and where the ROI breaks down. Evidence from Stripe, Coinbase, Ramp, and Shopify.
A backend PR kept colliding with other merges on database evolution numbers. Four manual rebases later, I described the problem to Claude and let it write a routine to handle the rest.
Practical tips for getting the most from Claude Opus 4.7's 1M context window in Claude Code. Effort levels, proactive compaction, subagent delegation, and session management from daily production use.
30 minutes. No pitch deck. We'll talk through what you're working on and I'll tell you honestly whether I can help.
No pressure, no upsell. Just a conversation to see if there's fit.
Schedule an intro callInsights on AI integration, engineering leadership, and building production systems. Written from ongoing practice, not past experience.
Occasional emails. No fluff.