Automate annotation at scale. Drive auto annotation and auto labeling pipelines, reducing manual cost while improving label quality and throughput.
Build advanced data miners. Develop intelligent approaches to surfacing rare, surprising, and safety critical scenarios in very large datasets, using techniques such as embeddings, semantic search, learned representations, and model driven data selection.
Own log selection and storage. Define how we select, store, and retrieve fleet logs so the right data reaches our models efficiently and economically.
Connect data to model performance. Establish how we measure the value of data and make rigorous trade offs across quality, accuracy, speed, cost, and scale.
Close the loop. Feed curated data into model training and evaluation, then analyze outcomes to continuously refine what we collect and label.
- Partner cross functionally. Work closely with ML, metrics and evaluation, and infrastructure teams to keep the flywheel running reliably at scale.
- Qualifications
- Track record of building and leading high performing technical teams (data science, data engineering, ML, or labeling), including managing managers or senior individual contributors.
Deep technical expertise in computer vision, video, multimodal AI, or related perception problems.
Experience leading large scale data capabilities that directly influence model training, evaluation, and performance.
Strong intuition for what makes data valuable, with experience in data discovery, selection, curation, and enrichment at scale.
Proven execution at scale, with strong technical and commercial judgment across quality, speed, cost, and build versus buy trade offs.
- Ability to move between strategy and technical detail, set direction in ambiguity, and influence senior technical and business stakeholders.
- Bonus Qualifications
- Autonomous driving, robotics, or embodied AI.
Large scale video, multimodal, or foundation model training.
Semantic search, embeddings, active learning, auto labeling, or other approaches to intelligent data selection and enrichment.
Large scale real world data acquisition across fleets, partners, or multiple geographies.
