Building reliable
human-AI systems.
Design and build AI-native operating systems that help companies act on their data with precision and discipline — while teaching founders and engineers how to think in systems, manage risk, and build resilient organizations.
Build systems that think.
Teach operators who execute.
Most companies don't fail because they lack technology. They fail because they execute in the wrong order — optimizing tactics while the system is broken.
I've spent 20 years building systems that turn chaos into clarity — from $250B cash reconciliation at Amazon, to observability SDKs across 8 languages at Datadog, to AI-powered growth engines at Finsi.
My approach: engineer companies like high-reliability systems. Clear ownership. Feedback loops. Risk management. Automated decisions where humans add no value, human judgment where machines fall short.
After 20 years building systems,
these aren't theories. They're patterns.
Systems beat heroics every time
Repeatable processes compound. Heroic effort burns out. I build infrastructure — technical and organizational — that lets companies scale without breaking the founder.
AI-native means rethinking the whole stack
Most companies bolt AI onto broken processes. Real transformation means rebuilding the decision loop: data → insight → action, with agents doing the work humans shouldn't.
Complexity isn't maturity. Simplicity is resilience.
The best systems look deceptively simple. They're not — they've earned their simplicity through rigorous thinking about failure modes, dependencies, and operator load.
If everything is under control, you're moving too slowly
Risk management isn't about avoiding risk. It's about knowing which risks to take, which to mitigate, and which to accept — then building systems to handle the fallout.
Here's what breaks most engineering organizations:
You build. Traction happens. You scale the team. Growth plateaus. So you add more: more engineers, more tools, more processes, another framework.
You're doing "all the right things" and everything gets slower, not faster.
I don't teach you everything. I show you what to fix next.
Think in systems, not features
Most technical organizations optimize the wrong things. I help teams see the system — constraints, feedback loops, second-order effects — and fix the actual bottleneck.
Build AI that ships, not AI that demos
The gap from AI pilot to production is 5-10x in cost and complexity. I've crossed that gap at AWS, Datadog, and now at Finsi. I know where the bodies are buried.
Design teams that operate like systems
Clear ownership, operating cadence, feedback loops, accountability. Not process for process' sake — structure that makes the team 10x more reliable.
Turn data into decisions, not dashboards
Dashboards are where insights go to die. I build execution layers: data → insight → automated action. The system decides, the operator validates.
Companies, systems, outcomes
Finsi
CEO & Co-FounderBuilding AI agents that connect siloed systems into a single decision and execution layer for omnichannel brands. Secured beta customers and agency partnerships within 3 months of MVP.
Datadog
Senior Engineering ManagerLed APM client libraries across 8 programming languages. Introduced adaptive sampling reducing customer costs by 10-30%. Supported $40M+ in pre-sales deals.
Amazon / AWS
Software Development ManagerDelivered AWS SDK for Rust, S3 client with 90Gbps throughput, and scaled Amazon's world-class cash reconciliation system processing millions of daily transactions.
DataArt → Luxoft
Director / Engineering ManagerBuilt the first subscription-based KYC system for 12 of the G14 investment banks. Led organizations from 30 to 110 engineers across 5 countries.
Ready to build with clarity?
Whether you're scaling a team, launching an AI initiative, or figuring out what to fix next — let's talk systems.