{"id":72757,"title":"Computer Vision Engineer – Video & VLM (m/f/d)","company":"Almetra","city":"Berlin","region":"Berlin","location":"Berlin, Berlin, Berlin, Germany","work_model":"hybrid","employment_type":"full_time","employment":["full_time"],"roles":["data_science"],"visa":"not_mentioned","relocation":"not_mentioned","posted_at":"2025-05-07T11:37:07.086000+00:00","tools":["Computer Vision","PyTorch","Python","Docker","AWS"],"seniority":"unknown","years_experience_min":null,"german_requirement":"not_mentioned","german_level":"unspecified","english_requirement":"required","english_level":"unspecified","team_language":"unknown","confidence":0.4,"method":"rules","language_badge":"Not stated","language_evidence":null,"requirements_snippet":"Person detection and tracking in crowded scenes, especially transformer-based detectors (DETR family, Grounding DINO) TensorRT, quantization and inference optimisation for edge Temporal grounding or video captioning Publications …","url":"https://jobs.ashbyhq.com/almetra/5ede6aaa-24e6-4b35-8966-369b052c2b1b","source":"ashby","agency":null,"description_text":"At Almetra, we’re on a mission to reinvent how factories work.\nEvery car, wind turbine and medical device comes from a factory. But factories are running out of people and need better tools to see what’s really happening on the shop floor.\nThat’s where we come in. Almetra builds AI-powered computer vision systems that show manufacturers how their production lines actually run, and how to improve them.\nAs a Computer Vision Engineer – Video & VLM, you’ll own the video models at the core of our product, from tracking people at each station to recognising every step of the work with vision-language models. You’ll work closely with our Head of VLM, the ML team and our platform engineers.\n\nYour role\n• Improve our computer vision pipeline end to end, from person and hand tracking to action recognition with VLMs\n• Fine-tune and evaluate vision-language models on real factory video, especially on steps that look alike\n• Run models reliably on video in the cloud and on edge devices in factories, with checks that catch weak results early\n• Make models faster and cheaper to run (distillation, quantization, smaller backbones)\n• Help shape how we approach video understanding at Almetra\n\nYou are\n• Experience building computer vision and deep learning models for video (not only still images), ideally used by real users\n• Hands-on experience fine-tuning VLMs or large video models (e.g. InternVL, Qwen-VL or similar)\n• Strong PyTorch and multi-GPU training, plus clean, reproducible Python\n• Experience running models in production pipelines, in the cloud and/or on edge devices (Docker, AWS, Jetson or similar)\n• A pragmatic mindset: you care that it works every day, not only on the benchmark\n\nNice to have\n• Person detection and tracking in crowded scenes, especially transformer-based detectors (DETR family, Grounding DINO)\n• TensorRT, quantization and inference optimisation for edge\n• Temporal grounding or video captioning\n• Publications on video or multimodal learning\n• Industrial, retail or other real-world camera settings\n\nWhy this role matters\n\nYou bridge research and production — turning cutting-edge computer vision models into robust, real-time systems that power intelligent industrial automation.\nWhat we offer\nEmployee Share Options Program for all permanent employees*\nAn increasing benefits list: currently includes Urban Sports club and quarterly team retreats.\nBe on the forefront in defining what artificial intelligence means in manufacturing\nGain hands-on experience in working in an AI-first software company\nSupportive and inclusive culture that values diversity and promotes the advancement of underrepresented groups within the company\nCollaborate with a diverse (currently more than 10 nationalities) and talented team, working on cutting-edge projects with real-world impact\nNetwork with professionals and leaders in the field, opening doors to potential future career opportunities\nWe have a very flat hierarchy, open 360° feedback, and flexible working hours\nEthics⚖: We are committed to developing ethical AI software\n\nDon't meet all the requirements?\nAlmetra is committed to creating a workplace that is diverse, fair, and inclusive. We encourage candidates from all backgrounds, even if they do not meet every qualification, to submit their application. We firmly believe that having a team with diverse perspectives only strengthens our company and drives innovation. Our commitment also extends to providing an accessible environment for everyone, including those with disabilities. Please let us know if you require any accommodations during the application process or while working with us, and we will do our best to support you.\n\n*Only full-time, permanent roles are eligible for stock options. Part-time roles, contract roles, work-student, internships and freelance roles are not eligible for stock options.","responsibilities_text":"• Improve our computer vision pipeline end to end, from person and hand tracking to action recognition with VLMs\n• Fine-tune and evaluate vision-language models on real factory video, especially on steps that look alike\n• Run models reliably on video in the cloud and on edge devices in factories, with checks that catch weak results early\n• Make models faster and cheaper to run (distillation, quantization, smaller backbones)\n• Help shape how we approach video understanding at Almetra\nYou are\n• Experience building computer vision and deep learning models for video (not only still images), ideally used by real users\n• Hands-on experience fine-tuning VLMs or large video models (e.g. InternVL, Qwen-VL or similar)\n• Strong PyTorch and multi-GPU training, plus clean, reproducible Python\n• Experience running models in production pipelines, in the cloud and/or on edge devices (Docker, AWS, Jetson or similar)\n• A pragmatic mindset: you care that it works every day, not only on the benchmark","requirements_text":"• Person detection and tracking in crowded scenes, especially transformer-based detectors (DETR family, Grounding DINO)\n• TensorRT, quantization and inference optimisation for edge\n• Temporal grounding or video captioning\n• Publications on video or multimodal learning\n• Industrial, retail or other real-world camera settings\nWhy this role matters\nYou bridge research and production — turning cutting-edge computer vision models into robust, real-time systems that power intelligent industrial automation.","status":"open","duplicate_of":null,"first_seen":"2026-11-02T21:26:53.761271+00:00","responsibilities_summary":null,"requirements_summary":null,"all_tools":["Computer Vision","PyTorch","Python","Docker","AWS"],"posting_language":"en","evidence":[],"years_experience_max":null,"mobility_evidence":[],"model_version":"rules-8","english_badge":"English required","company_id":364}