Hi, I'm Arash

I've spent close to two decades building software that runs at scale — cloud platforms, distributed systems, and lately, AI orchestration. Along the way I've learned that the best systems aren't just technically elegant; they're built by people who trust each other, and they exist to serve a real business need.

I believe in people. I believe in giving chances, motivating, and mentoring. Some of my proudest moments have nothing to do with uptime numbers — they're about watching engineers I've worked with grow into leaders themselves.

What I care about

I'm drawn to hard problems at the intersection of scale and reliability. Over the years I've led migrations of hundreds of services and databases across AWS, GCP, and Azure — the kind of work where "zero downtime" isn't a slogan, it's the acceptance criteria. I've helped re-architect monolithic platforms into distributed microservices, and I've seen firsthand how the right architecture can turn linear growth into exponential scaling.

Lately, my curiosity has pulled me toward AI orchestration: agentic workflows, Model Context Protocol, vector databases, and the pipelines that feed them. It feels like the early days of cloud computing again — full of potential, short on convention.

I also care deeply about the unglamorous stuff: observability, incident response, compliance. I've translated frameworks like FedRAMP, SOC 2, and PCI DSS into engineering reality, and I've watched teams go from firefighting to a 99.99% availability culture. Reliability is only truly tested when incidents happen — and how a team responds says everything about it.

How I work

I lead by example and by context, not by title. I've worked closely with C-level executives to turn long-term vision into achievable milestones, and just as closely with engineers to make those milestones real. I believe complexity must never be a design feature — it must only exist to justify a need.

Some things I've learned along the way:

• Small, focused tools beat sprawling platforms — in code and in process
• Self-service is a culture, not a tool; automation that covers 95% of manual steps changes how teams think
• Performance work compounds: a 25x improvement in one hot path can save millions a year
• Compliance, done right, is a product feature — it opens markets instead of closing doors

What I work with

My day-to-day toolkit spans Kubernetes, serverless, CI/CD, and the major clouds. I'm comfortable deep in PostgreSQL, MySQL, Redis, Elasticsearch, and the rest of the data layer. I've written production code in C#, Java, JavaScript, Python, and more — though these days I spend more time designing systems than writing them, and I still love reviewing code.

On the AI side: agentic workflows, MCP, LLMOps, and vector databases like pgvector.

Writing

I write about the things I experiment with — distributed systems, networking, databases, and the occasional home-lab adventure.

• This TCP Congestion Algorithm Could Speed Up Your High-Latency Links (with a twist) — Aug 2025
• Building a Distributed Counter with Raft and Gossip: My Experiment With Consensus and Cluster Coordination — Apr 2025
• Enabling Read-Only Mode in PostgreSQL Using Extensions — Nov 2024

If any of this resonates, say hello.