The posting
About Perceptyne:
Perceptyne Robots is a deep-tech robotics company building highly dexterous, low-cost, and contextually intelligent robots to automate tasks that largely remain elusive to robots, even today! These robots can be used in areas like electronics product assembly, e-commerce packaging, food preparation, etc. We are building technology that will push the boundary of robotics as we know it today! (www.perceptyne.com)
How is it like to work at Perceptyne?
At our core, we like to build things, we evangelize great ideas, and we are grounded.
At Perceptyne, we greatly value these dichotomies:
- Having experience while not losing a first principles approach.
- Having a large vision while not losing attention to detail
- Working on disruptive tech while hitting the market ASAP
Work Description:
As an AI/ML Engineer (Robotics Research), you will be involved in building the intelligence of the robot, end-to-end.
You will focus on leveraging machine learning to marry multi-sensor perception, model prediction and robotic actuation. You will come up with innovative methods / pipelines, bootstrap them in simulation and deploy them to control real robot behavior.
You will come-up with novel ways to engineer, capture data and train models using various sources including real-world behaviors, active planning, feedback-based control, demonstration-based learning, simulation and transfer learning, and real-world adaptation.
Your everyday activities shall include:
- Innovate to come up with smart, new methodologies in robotic perception and control
- Play a key role in end-to-end AI/ML system architecture design
- Theoretical research and hands-on implementation of new AI models, benchmarking and iterating them
- Building Computer Vision modules for robotics applications
- Explore and Identify suitable SOTA models, test/benchmark them for our applications, modify / fine-tune / improve them
- Build model pre-processing and post-processing pipelines in Python / C/C++
- Optimize the stack for real-time execution with focus on efficient programming and faster execution times
- Prototyping & testing concepts, ideas & hypotheses
- Testing various schemas in perception, planning, manipulation & control
- Testing & deploying production ready software
Skills & Requirements:
- Experience Requirements:
- Overall experience ranging from 4 to 8 years in software development.
- At least 4 years of industry experience as an FTE in building, training and testing AI/ML models and Computer Vision applications
- Experience building end-to-end system architecture with focus on execution speed and efficiency
- Bachelor's in computer science / Electrical / Electronics / Communications Engineering
- Masters / PhD in fields related to Computer Vision / Robotics / AI is an added advantage
- Technical skill Requirements
- Excellent understanding of architectures in AI, including Transformers, CNNs, RL, Diffusion Policies, Behavioral Cloning, Vision-Action models, VLMs etc. along with their respective applications
- Hands-on with computer vision with strong focus on building edge-deployable high-framerate systems
- Strong coding skills in Python and C/C++ and experience with PyTorch/Tensorflow
- Prior experience in robotic applications is an added advantage
- Experience with GPU optimization for edge-deployable applications is a big plus
- Strong aptitude towards debugging and root-cause analysis
- Strong intuition for relevant math & physics, coordinate geometry, trigonometry, calculus, linear algebra
- Good understanding of state-of-the-art robotics research
- Good theoretical / mathematical understanding of robotic manipulators (Forward & Inverse kinematics, Jacobian, DH representations, path planning, control schema, etc.) is an added advantage.
- Soft skill Requirements
- Passionate about Robotics & AI.
- Be resourceful and creative about getting things done with a sense of urgency.
- Go-getter attitude
- Experience in building / guiding a team will be an added advantage
How to apply?
- If you are excited by what we’re building and would like to join us, apply here, or write to us at [email protected]
