Because what we build actually goes into the water and does the job - for both commercial and military use. Its real-world autonomy, not just theory.
We’re a driven and focused team. You’ll have space to try things, ask questions, suggest better ways, and be heard. Your work won’t disappear into a backlog. It’ll shape the product. We care about doing things well, learning as we go, and having each other’s back.
If you want to work on something that matters, with people who actually care, this might be your place.
If all that resonates with you - send us your resume. We’d love to chat and see if this is a great fit for you.
responsibilities :
Fine-tune and adapt open-weight LLMs (e.g., LLaMA, Mistral, Phi, GPT-J) for behavior reasoning and decision-making tasks.
Develop training datasets based on real-world mission logs, vessel behaviors, and operator interactions.
Implement reinforcement learning (human-in-the-loop or simulated) to improve decision quality.
Integrate LLM components into the autonomy software stack, collaborating with perception, planning, and control teams.
Evaluate model performance using simulations and behavioral benchmarks.
Build pipelines for data preprocessing, labeling, and continuous improvement.
Ensure safe, interpretable, and verifiable model outputs.
Support the deployment of optimized models in edge or onboard environments.
requirements-expected :
Bachelor’s or Master’s degree in Computer Science, Machine Learning, Robotics, or related fields.
3+ years of hands-on experience with deep learning and transformer-based models.
Strong knowledge of Hugging Face Transformers, PyTorch, or JAX.
Proven experience in fine-tuning and deploying LLMs (LoRA, PEFT, or other parameter-efficient techniques).
Practical knowledge of TensorFlow, CUDA, and CUDNN.
Familiarity with behavior modeling, reinforcement learning, or autonomous decision systems.
Solid experience with data pipelines - from collection and curation to versioning and evaluation.
Strong C++ skills and experience integrating ML models into production software.
Experience with robotics or simulation tools (e.g., ROS, Gazebo) is a plus.
offered :
Opportunity to expand AI capabilities to include Large Language Models (LLMs).
Flexible work hours - with 3 mandatory days of work from Gdańsk office.
Office stocked with snacks and refreshments.
Sports & wellness allowance - We support an active lifestyle with subsidies for gym membership.
Eyewear subsidy - We’ve got your vision covered with financial support for corrective glasses.
