Software engineer and technical leader with 14 years of building products, the last six as co-founder and CTO of Outfindo, an AI product-data startup. I designed its platform from scratch, led it through three pivots and four funding rounds, and brought LLMs into production in early 2023, long before it was easy. Before that I grew from developer to CTO at Dáme jídlo (Delivery Hero), scaling it to 1M orders a month and leading its migration onto the global platform.
I'm at my best hands-on: designing systems, prototyping, and turning experiments into reliable products, with a small team around me. Looking for a staff / principal or tech lead role where I can build again.
// how I work
Build things properly — and know when not to. A PoC answers a question and is fine to throw away. Stable costs ten times more, and that's correct. Pretending one is the other is what bites.
Experiments first. Small test, see if it works, scale what works, drop the rest. Most of our big tech leaps started as my one-person experiment.
Documentation is the team's memory. With AI agents more than ever: what's in someone's head should be on paper.
About AI I'm neither a sceptic nor a blind optimist. I'm a developer. I want things under control, but done efficiently — knowing when to let the agent write it and when to read the code line by line.
Honest teams. Say things straight, respect across seniority, no assholes. I'm not the smartest person in the room and don't need to be.
// experience
Outfindo — Co-founder & CTO
Aug 2020 – Sep 2026
From a B2C bike marketplace to AI product data and content for e-commerce (Heureka Group and others). Board member.
Joined as the first engineer and built the platform almost from scratch: product catalog and data model (still in production after 6 years), PHP/Symfony backend, REST API, SPA frontend, real-time filters on Elasticsearch, early switch MySQL → PostgreSQL. Main author of the core codebase the whole time.
Stood at the birth of every big tech leap, mostly as my own experiment first, then led it into production. Many of them directly carried a funding round.
Spring 2023: when GPT-4 came out, tried whether an LLM can extract structured product data instead of brittle scrapers. It could. This became Diver, our product-data engine, and part of the pitch for the EUR 1M seed (Presto Ventures). Brought in and drove Beacon, automated consumer-insight research (NLP over reviews and articles + LLM synthesis).
Led the pivot from B2C marketplace to B2B SaaS in a few months — and then pushed through switching the marketplace off. Big cleanup, very satisfying deletions.
Built product-data delivery for Heureka Group: async processing, queues, parallel workers, up to 30k products a day; several hundred categories and over a million products delivered, in multiple languages. Built quality control into the pipeline: sourcing from multiple sources, fact checks per item, automated QA loop.
Owned the tech side of all four funding rounds: strategy, architecture defense, due diligence, pitching with investors (incl. Eleven Ventures).
Built the product & engineering team from zero to ~10. Hired through communities and honest job posts with salary ranges; bet on character over seniority — two juniors with almost no coding experience grew into owners of key processes. Team values, knowledge-sharing sessions, offsites, 1:1s.
2025–26: learned to work with AI agents (Claude Code) and evangelized it in the team. Set up the environment so agents work reliably: rules and documented processes in every repo, plan first, humans do irreversible and production actions. Then across systems: a central infra repo with decision records and service registry, documentation synced automatically from all services, MCP servers as interfaces for agents — and GUIs next to them, because processes and data must be visible for people to use them. Extracted the first service from the monolith and set up our first production Kubernetes deployment as a pilot; the team then migrated onto the cluster.
Delivery DJ (Delivery Hero) — Project & Engineering Lead, part-time
2020 – 2023
A small startup inside Delivery Hero: order dispatch for restaurants with their own couriers, with our own rider apps (iOS, Android) and vendor admin. Led 1–2 part-time developers, owned product decisions and onboarding of new countries.
Used by hundreds of restaurants in several European countries. Known for “just write us on Slack and we'll change it”.
Dáme jídlo (Delivery Hero, now foodora)
Jan 2016 – Jun 2020
Czech food delivery leader. Grew to ~1M orders a month by spring 2020.
CTOSep 2019 – Jun 2020
Member of management reporting to the CEO; ~20 people in product & engineering (backend, frontend, mobile, QA, product), responsible for the budget, hiring and raises.
Kept the platform running through the order growth and the first COVID wave, while shipping a ton of integrations and features. Key technical calls with the team: group orders, payment gateway, Datadog as one platform for monitoring and logging.
Set team values together with the team: we trust ourselves, we work together, we give each other space.
When Delivery Hero decided to move DJ onto its global platform, I led the team through the transition — openly, as a team, keeping the trust of people on both sides. Then six months of migration with Delivery Hero — on 1 July 2020 we switched over, turned off the lights and left. The team still meets almost every month, six years later.
Backend Team LeadJan – Aug 2019
Was frustrated and thinking of leaving; decided instead to face the problems. Proposed the team reorg myself and got the mandate. Planning, decision log, mentoring plans for juniors, workshops — and still ~50 % of my time coding.
Software Developer → SeniorJan 2016 – Dec 2018
Joined right after the Delivery Hero acquisition. Nobody led us at first — a small punk team that built its own culture: Scrum, retros (we called them “hejtro”), code reviews, pair programming, tests, coding standard. Finally building things properly.
The whole PHP monolith: orders, payments, queues and jobs; transactions, concurrency, locks under growing traffic. Moved the architecture towards DDD / hexagonal with our own message, command and event bus (open-sourced, incl. a PhpStorm plugin). Split the frontend into an SPA, moved to GCP with autoscaling, monitoring and alerting.
System for processing legal information requests for major Czech banks: ingesting requests from PDFs, data boxes and OCR, integration with the banks' core systems. On-site installations, go-lives, maintenance. PHP on Oracle and MSSQL with lots of logic in stored procedures. Legacy code, a fork for every client — I learned what I don't want, and went looking for how to do it right.
Earlier
2001 – 2012
Web design since ~2000. Information system for Honzík language school (Jihlava): from requirements to maintenance, PHP/MySQL, at least two big iterations — tutors, lessons, attendance, the whole agenda. To my surprise it ran for years without me; in 2019 I helped migrate its data. Subject of my bachelor's thesis.
AI & data: LLM extraction and content generation in production, QA and evaluation loops, LLM cost control, MCP servers; Python data stack with the data team
Leading: building teams from scratch, hiring, 1:1s, team values, budgets, board and investor reporting, tech due diligence
Ways of working: AI-assisted development (Claude Code), ADRs, experiments, code review, pair programming
// open source
Older PHP projects from 2015–2020, mostly for historical interest: php7backport (backports PHP 7 code to PHP 5, it caught on and helped me get the job at Dáme jídlo), message-bus with Nette integration and a PhpStorm plugin, coding-standard. Work since 2020 is in private company repos. → github
// education
Masaryk University, Brno — Bc., Faculty of Informatics (2012); Bc., Economic Information Systems, Faculty of Economics and Administration (2009)
// languages
Czech (native) · English (full professional) · German (limited)
// outside work
Making furniture, guitar, mixtapes on cassettes, retro games, trips with the family in our camper van.