Machine Learning Engineer | Robotics And AI

Embodied Intelligence Real-World ML / Applied Autonomy About the Company We are representing an expanding deep-tech startup focused on building cohesive ecosystems that bridge physical machinery and intelligent software. The company's core mission is to democratize advanced automation by engineering modular, out-of-the-box hardware and software suites that solve tangible challenges in the physical AI space. Role Responsibilities Model Development: Design, optimize, and evaluate sophisticated machine learning models tailored for robotic mobility and dexterity, leveraging both behavior-cloning and reward-based learning methodologies. Pipeline Ownership: Manage the end-to-end AI lifecycle, which includes orchestrating diverse, multi-sensor data streams, scaling up model training, and executing final deployments directly onto physical machines. Sim-to-Real Iteration: Drive rapid prototyping cycles between virtual environments and the physical world—gathering real-world telemetry, diagnosing edge cases, and continuously driving systemic improvements. Infrastructure Scaling: Architect and maintain robust, multi-node GPU training workflows, taking charge of resource allocation, experiment tracking, and model checkpointing. Tooling: Build custom internal utilities for deep-dive experiment reviews, visual diagnostics, and telemetry analysis. Cross-Disciplinary Collaboration: Work hand-in-hand with mechanical, electrical, and firmware engineering experts to ensure theoretical ML breakthroughs translate into reliable, field-ready automation. Core Qualifications Advanced degree (MSc or Doctorate) in Robotics, Artificial Intelligence, Computer Science, or a closely aligned field, backed by relevant industry tenure. Deep programmatic fluency in modern deep learning libraries, particularly PyTorch. Demonstrated mastery of physical AI techniques, specifically around policy learning, neural control systems, and agent-based training paradigms. A strong portfolio demonstrating the ability to move algorithms out of purely simulated academic benchmarks and successfully deploy them onto functional, physical hardware. Preferred Bonus Skills Practical exposure to large-scale, multimodal foundation models guiding physical agents (such as architectures integrating language, visual inputs, and kinematics), or generative video models. Experience handling high-throughput robotic data serialization formats, alongside a strong track record of contributing to premier academic venues (such as ICRA, IROS, NeurIPS, or equivalent). To apply online please use the 'apply' function, alternatively you may contact Evangeline. (EA: 94C3609/ R24124002)