Full-Time
Machine Learning / Reinforcement Learning Engineer
Eka
11-50 employees
Transport robotics platform with fleet management
No salary listed
Cambridge, MA, USA + 1 more
More locations: Boston, MA, USA
In Person
Bachelor's, Master's, PhD
 AI & Machine Learning (1) |
|---|
Get referred to Eka
See people who can refer or advise you
- A Bachelor of Science, Master of Science, or Doctor of Philosophy degree in Computer Science, Robotics, or a related field is required.
- Deep theoretical and practical knowledge of reinforcement learning and supervised learning algorithms is required.
- Experience with physics engines such as Isaac Sim, MuJoCo, or PyBullet and robotics middleware such as Robot Operating System or Robot Operating System 2 is required.
- A deep understanding of modern architectures, including Transformers, convolutional neural networks, and foundation models, is required.
- Expert-level Python skills and proficiency in deep learning frameworks such as PyTorch or JAX are required.
- A strong commitment to clean code, version control, and reproducible experimental workflows is required.
- A history of publications in top-tier robotics or machine learning conferences, or a portfolio of projects providing strong practical evidence of expertise in the field, is required.
- Research and implement reinforcement learning and supervised learning algorithms for robotic manipulation.
- Design simulation models and domain randomization strategies, and collaborate with the robotics team to ensure alignment with physical systems.
- Design experiments to evaluate and optimize model architectures for sample complexity and policy performance with real-time execution constraints.
- Develop scalable data management pipelines for real and synthetic data, and evaluate and select algorithms that maximize data efficiency and overall policy performance.
- Deploy, evaluate, and debug policies on physical hardware, identify bottlenecks, and implement improvements in collaboration with the robotics team.
Eka Robotics builds transport robotics systems that use automated guided vehicles, control software, and fleet management tools to move materials and automate handling. The platform trains robot controllers in simulation with self-supervised learning and then transfers the learned policies to real robots through sim-to-real reinforcement learning. It differentiates itself with a team from MIT, Berkeley, Harvard, DeepMind, and Boston Dynamics and a focus on physical AI, combining robotics, perception, control, and fleet management in one platform. The goal is to enable machines to automatically learn about their environment and continuously improve, delivering autonomous transport robotics and embodied intelligence for robotics and drones.
Company Size
11-50
Company Stage
N/A
Total Funding
N/A
Headquarters
Cambridge, Massachusetts
Founded
2025
Get referred to Eka
See people who can refer or advise you
Simplify's Take
What believers are saying
- Eka is hiring ML, robotics, and embedded engineers to scale R&D.
- Official site says it is deploying intelligence into the physical world.
- Investor backing from MFV Partners and E14 signals early conviction in the thesis.
What critics are saying
- Eka disclosed no customers, deployments, or commercialization timeline after launch.
- Funding remains opaque, despite conflicting $13 million and $2.5 million reports.
- VFA must beat VLA competitors like Physical Intelligence by benchmarks, or fade.
What makes Eka unique
- Eka Robotics launched April 29, 2026 with force-first VFA robotics.
- Pulkit Agrawal and Tuomas Haarnoja combine MIT and DeepMind research depth.
- Custom tactile grippers and sim-only reinforcement learning target superhuman manipulation.
Help us improve and share your feedback! Did you find this helpful?