Mission

Founding ML & Foundation Model Engineer

Greater Delhi Area (Hybrid) Full TimeFounding Engineer

Xolver AI is at the forefront of innovation, driving advancements at the intersection of AI, robotics, computer vision, and generative AI. Our mission is to design cutting-edge solutions that transform industries and make a meaningful impact on society. By leveraging state-of-the-art technologies, we develop groundbreaking products to redefine automation and intelligent systems, unlocking new possibilities for the future of Physical Intelligence.

Apply for This Role

Why this role matters

We are looking for researchers who love the "real world" and are obsessed with making models robust enough for messy, unpredictable environments.

Xolver / Infrastructure for Autonomy

Role Context

What this role actually is.

This is a founding engineering role for a specialist who will architect the "brain" of our physical intelligence platform. You won’t just be training models in a vacuum; you will be responsible for teaching machines how to perceive, reason, and act in unstructured environments. As a Founding ML Engineer, you will own the end-to-end pipeline—from managing massive multimodal datasets in the LeRobot ecosystem to training state-of-the-art Vision-Language-Action (VLA) architectures. You will bridge the gap between high-level cognitive reasoning and the low-level motor controls required for precise physical manipulation.

Scope of Work

The Actual Work.

Model Architecture & Training

Design and train large-scale foundation models and specialized policies (e.g., Diffusion Policy, ACT) that enable robots to perform complex tasks with high success rates.

Data Engine Ownership

Architect and manage the data pipeline for large-scale teleoperation and video data. You will leverage and extend the LeRobot ecosystem to curate high-quality demonstration datasets.

Sim-to-Real Evaluation

Develop rigorous evaluation frameworks in simulation (Isaac Sim, MuJoCo) to stress-test models before they are deployed to physical hardware.

Inference Optimization

Work closely with the systems team to ensure that massive neural networks can perform high-frequency inference on edge compute blocks without compromising safety.

Research Translation

Stay at the bleeding edge of robotics research, rapidly implementing and adapting new breakthroughs in VLA and generative physical AI for our platform.

Who tends to fit

You are a researcher who loves the "real world." You are frustrated by models that only work in papers and are obsessed with making them robust enough for messy, unpredictable environments. You have a "builder" mindset and aren't afraid to dive into raw data to understand why a model is failing a specific physical task.

What we expect

Core ML: Deep expertise in PyTorch or JAX, with a track record of training and deploying large-scale neural networks.
Robotics AI: Strong foundation in Imitation Learning, Reinforcement Learning, and Multi-modal perception.
Ecosystem Knowledge: Hands-on experience with Hugging Face’s LeRobot, and familiarity with datasets like Open X-Embodiment.
Vision & Spatial Reasoning: Deep understanding of 3D computer vision, spatial transformers, and temporal modeling.
Engineering Rigor: Proficiency in Python and experience with distributed training (DeepSpeed, FSDP) and ML Ops.
Preferred: A history of contributions to open-source AI or robotics research.
Preferred: Experience with Action Chunking with Transformers (ACT) or Diffusion Policies.
Preferred: Knowledge of Sim-to-Real transfer techniques and domain randomization.

Application

Ready to architect the
future of autonomy?

Share your CV at hello@xolver.ai.

Send Application