Edge AI Vision Engineer: Distill & Track Models

Harmattan AI; Zürich; On-site; CHF 120,000 - 180,000; Full time

Job summary

Harmattan AI is seeking an ML Research Engineer to join the Detect&Track Distillation team at an early stage, shaping the technical direction and distilling large foundation models into compact, task-specific components for edge and embedded systems.

You will work on target detection, classification, and re-identification, collaborating with the Detect&Track Foundation team and cross‑functional engineers in Lausanne, Paris, or Zurich to deliver robust, deployable solutions.

Qualifications

• Educational background in a STEM field (e.g., CS, Engineering, Mathematics).
• Proven experience in vision neural networks, target detection, or re-identification tasks.
• Hands-on expertise in knowledge distillation, model compression, and deploying networks on constrained embedded systems or edge hardware.
• Proficiency in MLOps, GPU compute, and building training pipelines and loggers.
• Highly structured, analytical, task-oriented, and capable of communicating benchmarking results to stakeholders.
• Commitment to Harmattan AI's mission and growth plans.

Responsibilities

• Model Distillation & Finetuning: distill foundation models into task-specific components.
• Edge AI Optimization: optimize networks for embedded systems using quantization, pruning, and LoRA.
• Pipeline Management & MLOps: build and manage training and evaluation pipelines with reproducibility and version control.
• Data Curation: assist in creating task-specific datasets to improve model accuracy.
• Benchmarking & Evaluation: benchmark distilled models for latency and real-world deployment.

Skills

Deep Learning

Computer Vision

Edge AI

Model compression

MLOps

GPU compute

Communication

Mentoring

Education

BSc in STEM

Tools

Jetson

NPUs

MLOps Pipelines

Job description

Harmattan AI is seeking an ML Research Engineer to join the Detect&Track Distillation team at an early stage, shaping the technical direction and distilling large foundation models into compact, task-specific components for edge and embedded systems.

You will work on target detection, classification, and re-identification, collaborating with the Detect&Track Foundation team and cross‑functional engineers in Lausanne, Paris, or Zurich to deliver robust, deployable solutions.