I work at the seam between GenAI research and shipped product. I own the data and vision pipelines, agentic workflows, and inference infrastructure that turn lab-grade techniques into something that real users love.
Four years on Marc Levoy's computational photography team at Adobe, shipping an iOS app and multiple Lightroom features used by tens of millions. Now at Phota Labs, working on personalizing generative photography to you. I think product-first, and build the tools I'd want to use myself.
I'm useful at two seams. Between research and product: taking something that works in a notebook and making it ship for a million people. And between engineering and product instinct: most ML engineers shipping at scale aren't thinking product-first; most people with product judgment can't ship the system underneath. Both together is rare, and it's why the work lands.
Most of my best engineering isn't visible from the outside, but you can see it in the products it ends up in. The parts that don't show are usually the parts that decide whether a feature actually survives in the real world.
Sometimes the engineering is turned into a publication; more often it just ships. I prefer to ship.
Building an identity-preserving generative photography product. I own the production stack — inference + serving infrastructure, computer-vision and data pipelines, and the API surface that turn the team's research into a shipped product. The launch post covers what's live now.
Multi-frame computational photography pipeline producing SLR-like raw and JPEG output on iPhone, from Adobe's Nextcam team. Core developer on the team.
A learned, scene-adaptive raw-rendering profile shipped in Camera Raw and Lightroom. I'm a co-author of the launch blog and built the training-data pipeline that shaped the model's architecture. Contributed to the productionization that took the technique from research prototype to shipped feature.
A one-click AI feature that removes reflections from photographs taken through plate-glass windows in Camera Raw and Lightroom. Scaled the real photo source training dataset from ~3k to >100k assets. Acknowledged contributor in the launch blog.
Liszt · Liebestraum No. 3
Stitched together from a few takes. Recorded with my phone at Brown Music Center at Stanford.
Beethoven · Piano Sonata No. 8 (“Pathétique”)
Beethoven is one of the few composers I never get tired of. Don't be discouraged by the slow start -- it turns into a really fun piece to listen to! Recorded with my phone at Brown Music Center at Stanford.
Piano arrangements of “Frozen” and “Demons” from years back →
I grew up in Germany, where I studied Electrical Engineering and Information Technology at RWTH Aachen. I then moved to California to study at Stanford with a focus on computational imaging and neural rendering. I've always been fascinated by building things, seeing how they work, and improving them. Following Stanford, I worked for a MicroLED startup that was acquired by Google. Then, I switched to Marc Levoy's computational photography team at Adobe. Now at Phota Labs, building personalized generative models and workflows, and shipping faster than ever!
Outside the engineering I've played classical piano since I was a kid, and I'm constantly chasing bigger waves and watching marine animals when kitesurfing. I like every aspect of nature, and ideally I can explore it with Lucy, the Siberian Husky.
Open to conversations about AI engineering, including infrastructure, data + inference pipelines, computational photography, generative visual systems, and image-quality work. If any of those overlap with what you're building, please get in touch.