Harmattan AI, a next-generation defense technology company, is seeking an AI Research Intern in Lausanne, CH, to work on deep learning for computer vision. The role involves model design, optimization, and research in edge deployment, with hybrid work and mentorship from experienced researchers.
Harmattan AI is a defense technology company developing autonomous and scalable systems for allied forces. After a $200M Series B funding round, the company is expanding its research teams to deliver mission-critical capabilities. The AI Research Intern will contribute to cutting-edge computer vision projects, guided by values of real-world impact, excellence, and tackling complex technical challenges in a rigorous, ownership-driven environment.
Responsibilities
- Develop and implement deep learning models for computer vision applications.
- Conduct research on CNNs and Vision Transformers, including experimentation with new architectures.
- Apply model compression methods such as knowledge distillation.
- Implement quantisation-aware training and post-training quantisation techniques.
- Explore network and dataset pruning to improve efficiency.
- Design lightweight architectures suitable for edge and embedded devices.
- Curate datasets, ensure balance, and address bias issues.
- Plan experiments, perform ablation studies, and ensure reproducibility.
- Evaluate models using metrics like mAP, IoU, and calibration.
- Analyze failure cases and test model robustness under data distribution shifts.
- Read, analyze, and implement research from top conferences including CVPR, ICCV, ICLR, and NeurIPS.
Requirements
- Strong foundation in deep learning, preferably using PyTorch.
- Experience with CNNs or Transformers through academic work or projects.
- Understanding of bias-variance trade-offs and model generalization.
- Familiarity with optimization principles and basic probability concepts.
- Interest or experience in model compression techniques.
- Enthusiasm for designing hardware-aware and efficient models.
Nice to have
- Experience with ONNX, TensorRT, TFLite, or LiteRT.
- Familiarity with experiment tracking tools such as W&B or MLflow.
- Prior experience conducting ablation studies.
- Exposure to dataset curation or annotation workflows.
- Previous involvement in research projects or conference submissions.
What the company offers
- Practical research experience in applied computer vision.
- Insight into model optimization and deployment on edge devices.
- Opportunity to implement and evaluate state-of-the-art research.
- Mentorship from senior researchers and engineers.
- Chance to contribute to academic publications or conference submissions.
About the company
Harmattan AI is a defense technology company focused on autonomous and scalable systems. The company emphasizes real-world impact, excellence, ambitious goals, and solving difficult technical problems. It operates in a high-performance environment that values rigor, ownership, and execution.
- Commitment to building technologies with real-world impact.
- Pursuit of excellence in all work.
- Setting and achieving ambitious technical goals.
- Tackling the most challenging technical problems.
- Operating in a rigorous, ownership-driven environment.
Quelle: öffentlich zugängliche Karriereseite des Arbeitgebers. Batchly ist nicht der Arbeitgeber und steht nicht notwendigerweise in einem Vertragsverhältnis mit dem Unternehmen.