Senior backend and DevOps engineer with 20+ years of Java and Spring. I take software from the first commit to stable production, and I build my own AI products end to end.
AI studio for YouTube storytellers: script, voices with real emotion, characters that stay the same in every scene, images, video and upload in one pipeline.
AI-powered YouTube channel analysis that reverse-engineers successful videos to uncover the patterns behind titles, thumbnails, hooks, scripts and audience response.
Fine-tuning centre for custom image models (LoRA) on serverless GPUs. The product backend never talks to a GPU vendor directly.
No more flying blind in Kafka: shows who uses which topics, what it costs, and replays messages safely without disturbing production.
The signals Western founders are missing from Asia: AI, EV, robotics and startup news from 30+ sources, scored for relevance by AI, translated and delivered as a daily report.
Every AsiafeedTech story goes out automatically, on schedule, to LinkedIn, X and the Buttondown and Beehiiv newsletters. One feed, four channels, no manual posting.
An AI livestream and content engine built to run 24/7: generates, renders and streams video around the clock.
The first fully automated video pipeline: script, voice, Whisper subtitles, FFmpeg render and upload to YouTube. The groundwork for Seedflows.
Image pipeline for epic fantasy videos: all visual assets of a story generated automatically through Google Imagen.
Reading live data straight from an e-bike's CANopen bus. Own PCB, designed and brought up with an ESP32, from schematic to running firmware.
Java and Spring Boot services with clean interfaces and an architecture that stays maintainable, from first design to production.
Docker, automated deployments, monitoring and log analysis. Software is only done when it runs reliably.
LLMs, voice, image and video models built into products with cost control, fallbacks and per-user billing.
Production keeps failing? I find the cause in logs, data and code, fix it, and make sure it stays fixed.
I've built and operated backend systems for more than 20 years, mostly with Java, Spring Boot and microservices. For me a working system includes automated deployments, monitoring, log analysis and calm debugging in production.
Under the name UmbrinCraft I build my own products, from the first idea to running in production: AI video tools, Kafka tooling, news automation. And when software isn't enough, I design the circuit board myself.
Pascal Clément is a Swiss freelance senior backend and DevOps engineer based near Bern. He has built backend systems for more than 20 years and runs his own software products under the name UmbrinCraft.
Yes. He takes on freelance projects in backend development, DevOps and AI integration, remote or on site in the Bern region, in German or English.
UmbrinCraft is Pascal Clément's studio. Its products include Seedflows.ai, PartitionPilot, AsiafeedTech and IteraMachina.
Mainly Java, Spring Boot, Angular, PostgreSQL, Kafka and Docker, plus AI models from Google, OpenAI, Anthropic and ElevenLabs, and hardware with ESP32 microcontrollers.
