Künstliche Intelligenz Ingenieur

Syn2core; München; Vor Ort; EUR 90.000 - 150.000; Vollzeit

Zusammenfassung

Syn2core is building AI Copilot on smart glasses to aid assembly-line workers in real time, processing video, audio and sensor data at the edge.

As a core ML engineer, you will own model building, training, and reproducibility, tackling synthetic data generation and spatial reasoning with SLAM, while productionizing audio and edge privacy features.

You will partner with the Lead ML Engineer and report to the CTO, shaping ML production culture with hands-on impact on real factory lines.

Qualifikationen

• Strong PyTorch experience with production ML workflows.
• Experience applying ML in real-time or edge environments.
• Familiarity with model optimization tools and deployment.
• Experience with SLAM or spatial perception in production is a plus.

Aufgaben

• Build the machinery to make models provable, including pipelines and registries.
• Develop synthetic data generation for rare anomalies.
• Advance production of audio modality and worker-vehicle association with SLAM.
• Ship at the edge and optimize for constrained GPU/CPU budgets.

Kenntnisse

PyTorch

Production ML

Edge ML

SLAM

Tools

ONNX

TensorRT

Jobbeschreibung

At Syn2core, we're building AI Copilot, an AI assistant that runs on smart glasses to support assembly-line workers in real time. Our system streams and processes data (video, audio, IMU sensor data) in real time, and delivers audio feedback directly to the worker. We're deploying it on the production lines of the automotive industry across the world - privacy-first by design: everything we persist is anonymized at ingestion.

Our long-term vision: the egocentric data we collect is training fuel for the next generation of robot learning - positioning us at the heart of the race towards autonomous robots for industrial deployment in automotive, aviation, aerospace and beyond.

The role

Our clients accept our system through formal tests with hard recall and false-positive gates - so our models must provably work: which weights, trained on which data, with which config, always answerable. You will own that machinery end-to-end, and with it two of our hardest ML problems: synthetic data generation for rare anomalies, and spatial reasoning - knowing which vehicle a worker is acting on as they move between cars, using SLAM fused with vehicle identity signals.

You'll work as a peer of our Lead ML Engineer - they own what the system should do, you own how models get built, trained, and reproduced - designing together, in the open, with a direct line to the CTO.

What you'll do

• Build the machinery that makes models provable - training pipelines, experiment tracking, model registry: full lineage from dataset to deployed weights, plus the evaluation harnesses we stake client acceptance on
• Solve data scarcity - simulation-based synthetic data pipelines for anomaly classes real factories are too good to produce often
• Take the system beyond vision - productionize our audio modality; develop worker-vehicle association with SLAM and vehicle identity signals
• Ship at the edge - own the anonymization models our privacy guarantees depend on; optimize everything for constrained GPU/CPU budgets on factory hardware

What you get

• The ML production culture of a company, shaped by you from the start - registry, tracking, evals, your way
• Multimodal problems (vision, audio, spatial) most teams only get one of, on data nobody else has
• Your models on real assembly lines at major OEMs within weeks, with measurable stakes

Where you'll be in 12 months

Every model that passes client acceptance is reproducible from the registry. A rare-anomaly class hit its recall gate on synthetic data. Audio is live in a deployment, and SLAM-based worker-vehicle association is validated on a real line. We'll get there together - the architecture with our Lead ML Engineer, the machinery yours.

Who you are

• You report the real number, especially when it's bad - our clients' acceptance tests leave no room for flattering evals
• You build machines that build models: reproducibility over heroics
• You prefer solving a problem once, generally, over solving it five times quickly
• You explore broadly, then converge and commit
• You're creative about data scarcity - synthesis, augmentation, simulation

Your experience

Must have:

• Strong PyTorch and production ML experience (detection / classification / tracking)
• SLAM / spatial perception used in production, not just coursework
• Model optimization for edge hardware (e.g. ONNX, TensorRT)

Ways to stand out:

• Manufacturing, robotics, or other physical-world domains