Director of Products at CodeClouds. 12 years building web products, cloud systems, and the teams behind them — started as a developer in 2014, been at CodeClouds since 2016 (junior → senior → team lead → PM → Director, same company). The infrastructure I've shipped handles 1M+ requests/day. The databases I've designed run at petabyte scale. Nine products built and in production — all still running.

I write about real decisions and real scars — the stuff that only shows up after you've lived with a system in production for a few years — at techunfiltered.dev.

On my own time I'm building Nothing Left, an open-source project tracker for teams and the AI agents working alongside them.

Numbers from real production, not side projects:

• 1M+ requests/day — infrastructure I designed and still maintain
• Petabyte-scale databases in production
• 30M+ records, sub-1s aggregations — MongoDB analytical queries on real tracking data, not single-record fetches
• 9 products shipped at CodeClouds — all in production
• Applied LLMs in production — CV parsing, candidate fit-scoring, and team workflows; not demos, actual shipped features

Outside CodeClouds, on my own time:

• Nothing Left — project tracking for people and agents. Issues, sprints, boards, time tracking, reports, and a knowledge base, plus an MCP server so AI agents work the same backlog humans do. AGPL-3.0, self-hostable with Docker Compose, free hosted beta. I run my own projects on it. Source →
• Vynko — Android budget and expense tracker. Shared household ledgers, offline-first, no bank linking, no SMS permissions, no ads. Google Play →
• iToolVerse — free developer and finance utilities, built whenever I need one.
• vaaky-highlighter — a WordPress plugin I've quietly maintained for 10+ years.

CodeClouds products, iToolVerse, and Vynko live in private repos. Nothing Left is the exception: fully open.

I don't write tutorials. I write about calls I had to make and still live with.

• Architecture & infrastructure decisions — the ones I got right, the ones I'd redo
• Systems that look fine in theory and break at scale — indexes, queues, rate limits, event-driven nightmares
• Engineering leadership without losing the craft — how I still read code, review PRs, and build tools while leading teams
• Builder stories — the decisions behind Nothing Left, Vynko, and iToolVerse, built after office hours

New posts ship every week at techunfiltered.dev.

• Reduce Docker Image Size: Measure First, Then Cut
• Why Your Docker Container Gets OOMKilled (Exit Code 137)
• SNS vs Kafka: When You Actually Need Pub/Sub vs an Event Log
• How We Reduced an AWS Bill by 40% Without Rewriting the Application

All articles →

Not a skills-list flex — just what's actually in my daily rotation at work and on side projects.

Languages

Frameworks & runtime

Data

Cloud & infra

Dev tooling