Harmattan AI
ML Research Engineer (Detect & Track Distillation)
1w ago
EuropeMiddlemachine learningmodel distillationcomputer visiontarget detectionclassificationre-identification
ML Research Engineer focused on distilling large foundation models into optimized, task-specific components for detection, classification, and target re-identification in autonomous defense systems.
About the company
- Harmattan AI is a next-generation defense prime building autonomous and scalable defense systems. Following the close of a $200M Series B, valuing the company at $1.4 billion, we are expanding our teams and capabilities to deliver mission-critical systems to allied forces.
- Our work is guided by clear values: building technologies with real-world impact, pursuing excellence in everything we do, setting ambitious goals, and taking on the hardest technical challenges. We operate in a demanding environment where rigor, ownership, and execution are expected.
Responsibilities
- Harmattan AI is heavily pushing the boundaries of autonomous systems, where the perception of the surrounding world through visual cues is a vital component. To make sense of incoming visual data and enable mission-critical downstream decisions, we have developed custom detection models. As our product and project portfolio expands, we are diversifying our efforts in this space across multiple embedded platforms.
- As an ML Research Engineer in the Detect&Track Distillation team, you will join us at a very early stage, giving you a unique opportunity to heavily influence the technical direction of the team. Operating out of Lausanne, Paris, or Zurich , you will focus on taking large foundation models and distilling them into highly optimized, task-specific components. Your work will span target detection, classification, and target re-identification across time, directly tackling the hardware inference constraints of diverse edge and embedded systems.
- Model Distillation & Finetuning : Take large foundation models and compress/distill them into highly specific, efficient components optimized for smaller tasks and target detection.
- Edge AI Optimization : Optimize neural networks for constrained embedded systems using techniques such as quantization (PTQ vs. QAT), pruning, and LoRA.
- Pipeline Management & MLOps : Build, heavily modify, and manage training, evaluation, and MLOps pipelines while ensuring reproducibility, robust logging, and version control.
- Data Curation : Collaborate on data curation and the creation of task-specific datasets to constantly improve model accuracy.
- Benchmarking & Evaluation : Framework-level benchmarking of newly distilled models to evaluate performance and latency, ensuring results are fully aligned with real-world operational deployments.
- Research & Innovation : Stay at the absolute forefront of scientific trends in computer vision and quantization research to introduce cutting-edge methodologies to the team.
- Cross-Functional Collaboration : Work closely with the Detect&Track Foundation team, downstream System Engineers, Project Teams, and Mission Intelligence to deliver robust solutions.
- Mentorship : Depending on seniority, support the team by managing or mentoring junior engineers.
Requirements
- Educational Background : A strong academic record with a degree in a STEM field (e.g., Computer Science, Engineering, Mathematics).
- Deep Learning & Computer Vision : Proven experience running vision neural networks, developing target detection architectures, or managing re-identification tasks.
- Model Compression & Edge AI : Hands-on expertise in knowledge distillation, model compression, and deploying networks onto highly constrained embedded systems or edge hardware (e.g., Jetson, custom NPUs, wearables).
- Technical Competence & Infrastructure : Proficiency in MLOps, GPU compute, and building infrastructure (such as training pipeline templates and loggers).
- Professional Attributes : Highly structured, analytical, task-aligned, and research-oriented.
- Excellent communication and influence skills, with the ability to effectively translate and present complex benchmarking data to downstream users and senior stakeholders.
- Thrives under pressure in a fast-paced environment with a “no-task-is-too-small” mentality toward building foundational team infrastructure.
- Commitment : 100% dedication to Harmattan AI’s mission, vision, and ambitious growth plans, ready to go the extra mile to ensure operational excellence
- We look forward to hearing how you can help shape the future of autonomous defense systems at Harmattan AI.