I fix the infrastructure nobody owns.
Startups build fast, with developers and now with AI. The result works until it doesn't: deploys people fear, one engineer holding everything in their head, an AI feature whose bill and behavior nobody can explain. Every engagement starts with what you have already shipped.
services
Diagnose and fix2-4 weeks
Starts with a 1-2 week diagnostic: written assessment, prioritized risk register, a 90-day plan. Then I implement the highest-value fixes from it - cloud and AI spend, reliability, deploys - as a scoped second phase. You can stop after the diagnostic; most of the value is in the fixes.
Good fit if the system was built fast, the person who knew it best is gone or overloaded, and you want it examined and repaired by the same senior hands.
Stabilization retainermonthly
Ongoing senior ownership of the production side: infrastructure, deploys, cost, agent and prompt tuning, evals, on-call for AI-system issues. Bounded monthly hours.
Good fit if you shipped an AI system and need a senior engineer on call who actually understands the production path.
Every engagement is fixed price, scoped on a call: a defined end, named deliverables, and you own everything I build - code, runbooks, and the knowledge to run it without me.
Something else in mind - part-time, embedded, a scoped build? That works too: the shape gets set on the call. The one thing I don't sell is a report that confirms everything is fine.
Get in touch
recent work
Data platform for a US marketplace startupMar to Jul 2026, solo
Took over a stalled data initiative and shipped it to production in six weeks, one engineer across the whole stack.
- Domain-keyed warehouse unifying nine data sources across email, calls, LinkedIn outreach, and direct mail: 425k cross-source merges, zero orphans.
- Go API and a self-serve ingestion UI the growth team uses without engineering in the loop. Terraform and AWS underneath.
- Built to run without its builder: scheduled, watermark-resuming pipelines operated unattended and kept running after a clean handover.
roles
Solo infrastructure engineer, Gasp / Mangata Finance
Sep 2022 to 2026
$3M peak TVL, 22 months zero outages across 25 GKE nodes and 100+ microservices. Reference: Peter Kris, CEO, on request.
DevOps to CTO, Lavarage (Pine Protocol)
Jun 2022 to Jun 2023
Promoted to CTO after 6 months. Launched a DeFi protocol mainnet 30 days from inception; $2.8M peak TVL.
Senior platform engineer, NinjaHoldings
Aug 2021 to Jun 2022
US fintech. Migrated 30+ Python services EC2 to EKS; API p95 2.1s to 340ms.
Sole DevOps to SRE team lead, Takeoff Technologies
Mar 2018 to Jul 2020
Walmart and Costco warehouse automation across 8 isolated client environments; 99.95% uptime.
Also deployed on Solana mainnet: payment splitting, deadline validation, agent identity (ERC-8004).