# Rafał Leszko

> Staff AI / Backend Engineer at Daydream/Livepeer and lead developer on Scope — a real-time engine for interactive audio/video AI generation.

Based in Krakow, Poland. Working on real-time generative AI systems for audio and video, including video diffusion models, LTX-2, and ACEStep. Prior experience at Google, CERN, and Hazelcast. Speaker at 70+ conferences including KubeCon (twice). Author of "Continuous Delivery with Docker and Jenkins" (Packt Publishing, 3 editions).

## Key links

- [Homepage](https://www.rafalleszko.com/): Personal website
- [CV (PDF)](https://www.rafalleszko.com/cv/Rafal_Leszko_CV.pdf): Full professional CV
- [CV (JSON Resume)](https://www.rafalleszko.com/cv/resume.json): Machine-readable CV in JSON Resume schema
- [Portfolio](https://www.rafalleszko.com/portfolio): Selected projects (Scope, go-livepeer, Catalyst)
- [Scope](https://github.com/daydreamlive/scope): Real-time audio/video AI generation engine (lead developer)
- [GitHub](https://github.com/leszko): Open-source work
- [LinkedIn](https://www.linkedin.com/in/rafal-leszko/): Professional profile
- [Blog](https://www.rafalleszko.com/blog): Technical writing
- [Talks](https://www.rafalleszko.com/talks): 70+ conference talks including KubeCon (twice)
- [Books](https://www.rafalleszko.com/books): Continuous Delivery with Docker and Jenkins

## Current focus

Real-time generative AI for interactive audio and video. On Scope: architecture, model research integration, and end-to-end performance work across pipelines like video diffusion, LTX-2, and ACEStep, running on NVIDIA RTX 4090/5090 (and H100/H200).

## Technology

Python, Go, PyTorch, CUDA, WebRTC, Diffusion Models, Real-time streaming, Low-latency systems, Kubernetes, Cloud-native infrastructure.

## Location

Krakow, Poland. Remote-first. EU timezone. Open to working with EU, UK, and US companies.