RLVR Engineer @ Quesma
Engineered Reinforcement Learning with Verifiable Rewards (RLVR) workflows for LLMs. Invented training data for state-of-the-art models and created benchmark tasks for the most recent industry priorities.
Staff Software Engineer with a founder’s perspective
I design and build production systems end-to-end - architecture, implementation, and long-term ownership. I know what it takes to ship from zero to exit because I’ve done it. Most recently: RLVR - Reinforcement Learning with Verifiable Rewards - the post-training method behind today’s frontier models.
I’m a Staff Software Engineer specializing in system design, integration architecture, and AI-driven platforms - SaaS, ML, and large-scale API systems built to last, not prototypes. I’ve also been on the founding side: co-founded get.chat and sold the company in 2024.
My AI work predates the LLM era. Today I work on both sides of the model: production automation on agentic AI and LLM APIs (Claude, GPT, Gemini), and RLVR post-training - designing training data and benchmark tasks that push state-of-the-art models.
I own architectural decisions end-to-end, from scoping to long-term maintenance. I work best when given real responsibility and a problem worth solving.
A track record built on shipping.
Engineered Reinforcement Learning with Verifiable Rewards (RLVR) workflows for LLMs. Invented training data for state-of-the-art models and created benchmark tasks for the most recent industry priorities.
Took ownership of product architecture and drove integration with the company tech stack. Designed and delivered the Partner API, enabling partners to self-provision Kubernetes-based WhatsApp Business inboxes and driving adoption for a holding company with 100+ service providers and 10k+ online stores. Defined architectural standards and best practices.
Co-founded and bootstrapped a WhatsApp Business Platform product from 0 to exit (acquired 2024). Designed the architecture, shipped the MVP to production, and scaled to hundreds of customers and dozens of partners.
Built a server that scales network I/O operations, based on a custom-made Docker Compose cluster management algorithm that dynamically allocates resources in an adaptive manner.
Delivered an ML-based recommendation system pipeline and a content management frontend for pre-installed software of a global mobile device manufacturer. Python, Django, HTML/CSS, Scikit-Learn.
Programmed web crawling and scraping software as microservices.
Built the company’s PoC and production business logic from scratch, enabling revenue growth. System engineering in Python, Docker, Django, PostgreSQL.
Languages, tools, and practices I work with every day.
Side projects that became real communities.
Niche video game series portal for mod creators and players in the Thief franchise. 5k unique monthly users, hundreds registered. Running and maintained since 2019.
A private Discord community of 850+ developers, founders, investors, and creators united around tech, innovation, and web3. Co-founded with Bogusz Pękalski in 2022 with the goal of building the most influential tech community in Poland.