About the job
Join Our Team as a Machine Learning Engineer
At Harmattan AI, we are pioneering the development of autonomous and scalable defense systems that make a meaningful impact. With our recent $200 million Series B funding, which has propelled our valuation to $1.4 billion, we are expanding our innovative teams to deliver essential systems to allied forces.
Our core values include a commitment to technological advancement with real-world applications, a relentless pursuit of excellence, ambitious goal-setting, and tackling the most complex technical challenges. We thrive in an environment where high standards, accountability, and execution are paramount.
Role Overview
As a Machine Learning Engineer on our Foundational team based in Paris, you will be crucial in developing the cognitive capabilities of our tactical robots. Your primary responsibility will be to architect and scale large, multi-modal foundational models that derive robust battlefield representations through Self-Supervised Learning (SSL) from extensive unlabelled Electro-Optical (EO) and Infrared (IR) data. These foundational weights will empower our Edge AI team to create highly accurate models for tactical hardware.
Key Responsibilities
- Architectural Design: Innovate neural network architectures (such as Vision Transformers) and loss functions (including Masked Autoencoders and Contrastive Learning) to effectively learn from both paired and unpaired EO and IR datasets.
- Training Infrastructure Management: Oversee and enhance training pipelines across multi-node GPU clusters, optimizing mixed-precision training and data loading processes.
- Evaluation of Representations: Establish metrics and linear-probing benchmarks to validate that the latent space captures meaningful semantic features prior to model distillation.
- Data Strategy Development: Assess current EO/IR data lakes and implement cross-attention mechanisms to integrate diverse sensor features.
- Collaboration: Work closely with Data Engineers on data ingestion pipelines and partner with the Edge AI team to ensure seamless transitions of high-performance models.
