Computer Vision Engineer – Object Detection (m/w/d)

Autonomous Teaming; München; Vor Ort; EUR 60.000 - 90.000; Vollzeit

Zusammenfassung

Autonomous Teaming is seeking a skilled individual in Munich, Germany, to work on computer vision and deep learning tasks related to object detection. The role requires building and optimizing neural networks, implementing perception algorithms, and collecting pertinent datasets. Candidates should have strong experience with computer vision models and proficiency in programming with PyTorch and Python. The position emphasizes collaborative success and ethical standards while requiring security clearance from NATO member or close ally nationals.

Qualifikationen

• Strong experience in computer vision and deep learning.
• Ability to modify and extend network architectures.
• Hands-on experience in dataset collection and labeling.

Aufgaben

• Build, adapt, and train neural networks for object detection.
• Implement perception algorithms utilizing temporal context.
• Collect, prepare, and curate real-world datasets.

Kenntnisse

Computer vision

Deep learning

Object detection

PyTorch

Python

Git

Docker

Tools

YOLO

FCOS

DETR

Faster R-CNN

RNNs

ConvLSTMs

Transformers

Jobbeschreibung

Your mission

• Build, adapt, and train neural networks for object detection and perception tasks using time-series data
• Implement perception algorithms that extract and leverage temporal context across multiple frames
• Design, implement, and optimize custom loss functions, augmentations, and architectures
• Collect, prepare, and curate datasets from the real world – not pre-cleaned academic datasets
• Supervise data annotation by refining labeling guidelines and assuring adherence to them
• Evaluate and debug models using AP, Precision/Recall, mIoU, and video-based metrics
• Deploy models into production as part of an integrated multi-sensor system (camera, radar, IR, acoustic, etc.)

Your profile

• Strong experience in computer vision and deep learning, with a focus on object detection using one-stage or two-stage models (YOLO, FCOS, DETR, Faster R-CNN, etc.)
• Strong experience with temporal models such as RNNs, ConvLSTMs, or transformer-based tracking
• Ability to modify and extend network architectures, not just use them off the shelf
• Hands-on experience in dataset collection, labeling, augmentation, and training
• Proven ability to train object detection models from scratch or adapt them to custom classes and data
• Familiarity with perception challenges specific to robotics, such as motion blur, lighting variation, occlusion, or limited compute
• Proficiency in PyTorch, Python, Git, Docker

Nice to have

• Experience with tracking algorithms (DeepSORT, ByteTrack, Kalman Filters)
• Prior work with non-visual sensors such as infrared, acoustic, radar, or LiDAR for object detection and fusion
• Exposure to model optimization and compression (quantization, pruning, distillation)
• Experience in the domain of UAVs, familiarity with ROS

What else

• Intrinsic motivation, attention to detail, and strong problem-solving mindset
• Creative thinker with a structured and systematic approach
• Humble and collaborative — prioritizes collective success over individual recognition
• High ethical standards and disciplined work ethic
• Fluent in English
• Nationality of a NATO member nation or close ally is required
• Successful candidates will be required to obtain security clearance