Founding Engineer

What we are building

We are a local-first desktop application that understands what happens on a screen. It captures real work as it unfolds, turning it to the company memory for both humans and agents.

(We are early and still mostly heads-down, so our page here is intentionally sparse. The full picture, meaning team and product, we are glad to share directly.)

The technical problem

Every environment runs on a different, messy stack of interfaces, so there is no clean API to integrate against.

That means learning to understand and segment what happens on screens and in video directly: parsing layout and state, detecting and segmenting the relevant regions and objects, recognizing actions and intent, and reasoning over long, noisy temporal sequences to find where one meaningful step ends and the next begins. Generic OCR and brittle per-app scripts fall apart here. Visual Language Models, combined with strong segmentation and temporal grounding, are the path forward. Building that layer well, meaning accurate, fast, and privacy-preserving on the user’s own device, is the core technical bet of the company.

We have a working proof of concept and real users in sight. Your job is to take the vision core from there to production, and to own it as it scales.

What you’ll do

• Own the CV/VLM core end to end, from raw screen and video capture to a semantic understanding of what is happening and why
• Build, fine-tune, and evaluate VLM and segmentation pipelines: spatial grounding, action and intent recognition, and temporal segmentation over long, noisy interaction sequences
• Make it run where it matters: on-device inference under real latency and resource constraints, built privacy-by-design and GDPR-native from the ground up. This is a hard constraint, and shaping the right approach to it is part of the role
• Define how we measure quality, including the benchmarks, failure cases, and standards that keep the vision layer honest as we scale
• Set the technical direction for vision and help shape the engineering team and roadmap as we grow

Who we’re looking for

• Senior or staff level computer vision engineer or researcher with genuine VLM and multimodal depth, from research or production
• Hands-on experience across the vision stack: segmentation, detection, video understanding, and modern VLM architectures
• You have shipped models into real products, not just notebooks, and you care about what happens after the demo: evaluation, edge cases, inference cost, maintainability
• You think in systems, not tickets. You want to own architecture, set the bar, and make the hard calls early
• You want the weight and the upside of a founding role, and you are ready to build in Munich (EU)

Why now, why us

This is a true founding seat. The founding team brings experience from SAP, Celonis, Tacto, and Enpal, so we know enterprise software, process intelligence, and how to go to market, and we are building at exactly the moment VLMs are becoming good enough to make this possible.

We are backed and supported by one of the most renowned labs for software and AI, which gives us the technical network, infrastructure, and credibility of an institution most startups never get access to at this stage.

Your salary is secured through state startup funding, so you get founding-level ownership and meaningful equity.

Targeted Full Compensation : €80–110k (Base + Equity) after stage funding.

Concrete numbers are based on experience etc., we are happy to discuss early and openly.

We strongly encourage everyone to apply, especially women and underrepresented candidates, even if you do not meet every requirement on this list.

This is the vision core of the company, and we are hiring the person who will own it. If that is the kind of problem you want to spend the next years on, we would love to talk.