Machine Learning Researcher

👋 About the Team

The Research team is Nota AI's research group based in Europe, focused on advancing model optimization for On-device AI and Physical AI.

Our goal is clear: to make Foundation Models such as VLMs and VLAs (Vision-Language-Action models) run in real time on actual devices and robots—not just on servers. We pursue state-of-the-art performance, with our research resulting not only in publications at leading conferences but also in technologies integrated into NetsPresso, our On-device ML framework, and ultimately into real-world products.

As a Machine Learning Researcher, you'll work closely with ML/DL Research Engineers, Frameworks Engineers, and Embedded Engineers, taking part in the entire process—from developing research ideas and publishing results to validating them on real hardware.

📌 What You’ll Do at This Position

• Do both research and product. Your research won't stop at conference publications. You'll see your work deployed on real customer devices through NetsPresso.
• Research at the intersection of two rapidly evolving fields. You'll work where Efficient AI meets Physical AI—an emerging research area focused on running Foundation Models in real time on robots and Edge devices.
• Small team, big ownership. As an early member of our European entity, you'll have the opportunity to take ownership of the entire process, from defining research directions to deployment, while collaborating globally with our R&D teams in Korea.

✅ Key Responsibilities

1. Physical AI Research & Development

• Research robot Foundation Models based on VLMs and VLAs
• Fine-tune VLA policies and evaluate their performance through simulation and benchmarking
• Explore and apply the latest research in Physical AI, including World Models and action-conditioned prediction

2. Model Optimization Research & Development

• Research optimization techniques for Foundation Models such as VLA, VLM, and LLM, including Quantization (PTQ/QAT, low-bit and mixed-precision), Pruning, and Knowledge Distillation
• Research training-free and test-time optimization, efficient attention, and efficient inference techniques such as KV cache optimization and token reduction

3. Real-World Validation & NetsPresso Productization

• Profile latency and memory and validate deployment across diverse target hardware, including GPUs, NPUs, robot arms, and humanoid robots
• Evaluate the downstream task performance of optimized models and contribute new capabilities to NetsPresso

4. Research & Publication

• Publish research at top-tier conferences such as NeurIPS, ICML, ICLR, CVPR, CoRL, and MLSys

✅ Requirements

• Ph.D. in Computer Science, Engineering, or a related field, or equivalent research experience
• A strong passion for independently identifying research problems and seeing them through to rigorous validation
• Deep research or development experience in at least one of the following areas:
• Deep learning model optimization, including Quantization, Pruning, and Distillation
• Training or fine-tuning Foundation Models such as VLMs, VLAs, and LLMs
• Professional experience with Python
• Strong proficiency in deep learning frameworks such as PyTorch
• Ability to investigate and analyze relevant prior research and independently reproduce published methods

✅ Preferred Requirements

• Publication record at top-tier conferences such as NeurIPS, ICML, ICLR, CVPR, ICCV/ECCV, CoRL, ICRA, or MLSys.
• Experience deploying and evaluating VLA models or robot policies in simulation or on physical hardware.
• Experience in robot learning, including imitation learning or reinforcement learning.
• Experience optimizing large-scale models such as LLMs or VLMs for model efficiency or inference/serving performance.
• Experience with On-device AI model conversion and deployment using TensorRT, ONNX Runtime, NPU SDKs, or similar technologies.
• Experience developing Triton or CUDA kernels.
• Experience developing with the ML ecosystem, including Hugging Face.
• Experience using AI coding tools such as Claude Code to automate research and experimentation workflows.