Senior, ML Engineer - VLM

About the role

Torc Robotics is on the lookout for a Senior Machine Learning Engineer to join our Vision-Language-Action (VLA) team. In this pivotal position, you will be instrumental in the creation and enhancement of datasets that are essential for our machine learning models, ensuring that our autonomous vehicle technology is powered by high-quality training data. This role presents a unique opportunity to collaborate with a passionate team dedicated to revolutionizing freight transportation through innovative software solutions.

Key facts

What you'll do

• Take the lead on the offline dataset pipeline, focusing on the design, implementation, testing, and deployment of cloud-based systems that transform multi-sensor data into training datasets tailored for VLM/VLA applications.
• Develop advanced VLM-assisted auto-labeling methodologies, such as open-vocabulary detection and dense captioning, to streamline the annotation process and minimize manual intervention.
• Create reasoning-grounded labels that are in sync with ego-motion and trajectory data, enhancing the training of VLA models and boosting their interpretability.
• Identify and curate complex scenarios to construct datasets that not only improve model performance but also add value beyond mere data volume.
• Oversee the data flywheel by defining dataset schemas and quality metrics, ensuring ongoing enhancement through iterative feedback loops.
• Work closely with the end-to-end model team to establish dataset specifications and uphold high standards for dataset delivery.
• Scale operations on cloud infrastructure by developing distributed pipelines that adhere to best practices in software engineering and documentation.
• Lead initiatives, mentor junior engineers, and ensure cohesive collaboration within the team and across the organization.
• Keep abreast of the latest developments in multimodal models and auto-labeling, applying research insights to real-world applications.

Requirements

• Demonstrated expertise in your field, capable of executing complex tasks with minimal oversight while exercising sound independent judgment.
• Expected to shape team dynamics and technical solutions, fostering consensus and providing mentorship to colleagues.
• A Bachelor's Degree in Computer Science, Robotics, Electrical Engineering, or a related discipline, with at least 6 years of relevant experience; OR a Master's Degree in a related field with a minimum of 3 years of experience.

Required Qualifications

• Strong background in Computer Vision and Deep Learning, with hands-on experience in model training and proficiency in at least two of the following areas: 2D/3D Object Detection, Tracking, Sensor Fusion, Semantic Segmentation, BEV, and Depth Estimation.
• Familiarity with multimodal or VLM applications, including experience with vision-language models, open-vocabulary recognition, dense captioning, or semantic embeddings related to perception data.
• Expertise in model data curation, particularly in developing datasets that enhance model performance, along with experience using large-scale data processing tools (e.g., Databricks, Daft, Pandas).
• Knowledge of distributed machine learning and data frameworks is essential.

Nice to have

• Experience with cloud computing platforms such as AWS, Azure, or Google Cloud.
• Familiarity with containerization technologies like Docker and orchestration tools such as Kubernetes.
• Previous involvement in research publications or contributions to open-source projects related to machine learning or computer vision.

Skills & tools

• Proficiency in programming languages such as Python and C++.
• Experience with machine learning frameworks like TensorFlow, PyTorch, or similar.
• Strong analytical skills and the ability to work with large datasets effectively.

Practical notes

This position is available to candidates who are authorized to work in the United States. Torc Robotics offers a competitive compensation package, including health insurance, retirement plans, and opportunities for professional growth and development. Interested candidates are encouraged to submit their resumes and cover letters by the specified deadline to be considered for this exciting opportunity.

About the company

Torc Robotics writes the software that lets heavy trucks drive themselves on public highways. The company began in Blacksburg, Virginia, in 2005 with students from Virginia Tech and later became part of Daimler Truck. Engineering and test work now run in the United States, Canada, and Germany, with an early commercial focus on Freightliner Cascadia trucks used in U.S. freight.