Jobs in Germany
About Us
STARK is a new kind of defence technology company revolutionizing the way autonomous systems are deployed across multiple domains. We design, develop and manufacture high-performance unmanned systems that are software-defined, mass-scalable, and cost-effective. This provides our operators with a decisive edge in highly contested environments.
We're focused on delivering deployable, high-performance systems — not future promises. In a time of rising threats, STARK is bolstering the technological edge of NATO Allies and their Partners to deter aggression and defend Europe — today.
About the team
The Data Operations team owns the entire data lifecycle behind STARK's AI stack: collection, acquisition, generation, curation, and management. We run our own data-collection campaigns across Europe, evaluate new sensors and platforms, and build the internal data platform that turns raw recordings into ready-to-use datasets. Everything we produce feeds directly into the perception and autonomy systems deployed on STARK's platforms — a real data advantage is built, not bought. The team is scaling up right now: real scope, direct impact, no legacy.
Your mission
Data is the fuel of STARK’s AI stack — you build the engine that makes it usable. You own the software backbone of our data platform: the metadata systems, ETL pipelines, data contracts, catalogs, databases, and internal tools that let engineers find, understand, validate, and reuse terabytes of multi-sensor field data in minutes, not days. You treat data context as a product: structured, searchable, version-aware, documented, and traceable from raw recording to processed asset, annotation delivery, dataset, and downstream ML workflow. Today, much of this is manual, scattered, or implicit — your job is to automate it away, support labeling efforts with the right data tooling, and turn operational data into reliable systems.
Responsibilities
Design, implement, and maintain our metadata database and data catalog (datasets, recordings, sensors, labels, lineage)
Build and operate ETL/ingest pipelines that bring field recordings, synthetic data, and external deliveries into our cloud storage (GCP)
Own the data management and labeling lifecycle end-to-end: coordinate and communicate with external labeling companies and data subcontractors, track deliveries, run QA reports, and build the operational workflows they work in
Develop internal enabling tools for the whole AI organization: dataset search and filtering, APIs/backend, dashboards, and self-service data access
Run data migrations and indexing jobs; keep the catalog consistent and fast as data volume grows
Handle admin support and user access management — and then automate these support tasks so they stop being manual work
Establish good engineering hygiene in a young codebase: tests, typing, docs, logging, CI/CD
Shape the long-term architecture and vision of the data platform together with the team
Qualifications
Strong Python
Solid SQL/PostgreSQL, including schema design
Experience with data modeling and metadata systems
Experience designing and operating ETL/data pipelines
Docker and CI/CD basics
Hands-on with object storage (GCS, S3, or similar)
Good software engineering hygiene: tests, docs, typing, logging
Organized and pragmatic: you can prioritize between a quick fix and a proper solution, and you know when each is right
Not allergic to support tasks — but technical enough to automate the support away
Comfortable coordinating with external vendors and non-technical stakeholders
Nice to have
Familiarity with ML datasets and labeling workflows (images, video, lidar; annotation formats like COCO)
Experience with synthetic data generation or GenAI-assisted data workflows (auto-labeling, data augmentation, foundation-model-based curation)
Experience with GCP services beyond storage (BigQuery, Cloud Run, IAM)
Experience with data versioning / dataset tooling (DVC, LakeFS, FiftyOne, or similar)
Experience in a startup environment — comfortable with ambiguity and changing priorities
Exposure to robotics data formats (ROS bags, MCAP, PX4 logs)
