Builder of systems
that matter.

Chapter 01 — Enterprise foundations

Learning systems thinking at enterprise scale.

My career began inside large enterprise software environments — the kind where reliability isn't aspirational, it's contractual. At SAP Ariba, I worked on procurement SaaS systems that thousands of enterprises depended on for mission-critical financial operations.

Enterprise scale teaches you things no startup can: the weight of accumulated technical decisions, the cost of brittle interfaces, and what it means to own a system with zero margin for downtime. I learned to think in systems — not features.

These years instilled a deep appreciation for operational excellence, platform thinking, and the kind of architecture that survives contact with reality.

Chapter 02 — The streaming era

Building infrastructure at the frontier of real-time data.

At Confluent — the world's data streaming platform, built by the original co-creators of Apache Kafka — I stepped into the center of the stream processing world. Confluent's mission is to put data in motion, and I was part of building the systems that made that possible at global scale across Confluent Cloud.

Working at a company where Kafka is the product — not just a tool — changed the way I think about data systems. Every architectural decision runs at 10x the stakes. Failure models, throughput guarantees, consumer group semantics: these aren't abstractions, they're what customers pay for.

I came away with deep convictions about how real-time infrastructure should be designed: not as a bolt-on, but as a first-class architectural primitive. Event-driven design isn't a pattern — it's a philosophy.

Chapter 03 — Startup & founding-level work

Building from zero with hard constraints.

At DeltaStream — the real-time context engine for AI agents, built on Apache Flink and ClickHouse — and other early-stage environments, the challenge flipped: you're building with a fraction of the resources, but the same quality bar. Founding-level infrastructure work demands clarity of thought, ruthless prioritization, and the ability to make 10-year architectural calls with incomplete information.

This is where BYOC architectures became central to my thinking. Building systems where enterprise customers deploy into their own cloud accounts — maintaining data sovereignty without fragmenting the product — is one of the hardest multi-tenant design problems in cloud software.

Startup infrastructure taught me velocity without sacrificing integrity. How to build abstractions that last. How to pick the right fights.

Chapter 04 — AI-native infrastructure

Building the infra layer for machine learning at scale.

At Kumo AI — a graph-native ML platform enabling relational deep learning over enterprise data graphs — I led the cloud engineering team building next-generation Kubernetes infrastructure across AWS, with planned expansion into Azure and GCP. The challenge: supporting SaaS, multi-tenant, and BYOC deployment models simultaneously while maintaining SOC2 compliance and driving cost engineering.

AI-native infrastructure is a different problem class than streaming infrastructure. The workloads are training-heavy with latency-sensitive inference paths, data access patterns are graph-traversal rather than sequential, and compliance (SOC2, data residency) must be baked into the platform architecture from day one — not bolted on later.

This experience reinforced a conviction: the engineers who can bridge streaming infrastructure, cloud platform engineering, and AI/ML workload requirements will define the next wave of intelligent systems architecture.

Chapter 05 — What I believe

Engineering as a discipline of consequence.

I believe the best infrastructure is invisible. When platform engineering is done right, product engineers don't think about it — they just ship. That's the goal. Platform teams exist to compress the gap between intent and production.

I believe distributed systems have a gravity that demands respect. They fail in subtle, cascading ways. They punish overconfidence and reward humility. Building them well requires holding the entire system in your head while also sweating the details of a single partition reassignment.

And I believe we're at an inflection point. AI-native systems are rewriting the infrastructure layer — new latency requirements, new data access patterns, new operational models. The engineers who understand both streaming infrastructure and AI workloads will define the next decade of cloud architecture.

That's the work I want to do. And that's the kind of founder or team I want to work alongside.