Execute the PCP metrics pipeline, assessing how hardware updates across various sensors impact PCP performance sensitivity in 3D scene understanding, live mapping and localization, especially in adverse weather and winter conditions.

Design and run structured A/B testing experiments, including side-by-side custom mount runs and tandem vehicle data collection, to clear sensors for the winter ODD.

Collaborate with AHE to correlate sensor-specific lab tests (e.g., heater/defroster impacts on camera lenses, snow accumulation on radomes) to downstream PCP performance.

Root cause detection regressions to understand how early fusion models compensate for modality degradation in challenging environments.

Design & develop performant and efficient multi-sensor fusion model to leverage the complementary advantage of on-board (low cost) sensors to improve Perception performance for highway and adverse weather.

Strong C++ and Python for running and modifying PCP metrics pipelines, automated testing, and re-simulations.

Experience with A/B testing methodology (RBV - Risk Balanced Verification) for evaluating hardware/software changes in varied environments.

Familiarity with diverse perception outputs and how environmental factors (like snow or fog) degrade bounding box precision/recall, point clouds, and 3D occupancy.

Ability to define complex data collection plans for winter ODDs (e.g., specific snow types, ice, low temperatures).

Experience with designing, developing and deploying multi-modality fusion based Perception models for 3D object detection, 3D segmentation & occupancy, Mapping and Localization. 

Experience in performance optimization to fit complex ML stack to low-power low cost edge compute (e.g., Nvidia Thor, TensorRT optimization).

Experience with designing better sensor or better ML model for an anomaly or rare event/objects detection

Translate  and coalesce company-wide feature requirements into concrete and comprehensive data deliverables

Work collaboratively with AI teams, project management, and data annotation teams to manage the collection and labeling of training and evaluation data powering Zoox’s AI perception stack. 

Manage vendor allocation and budgeting in conjunction with the milestone and release timelines at Zoox. 

Define and maintain clear tracking of outcomes, risk, and data quality to ensure transparency and accountability.

Design and train Vision-Language-Action (VLA) solutions for robotaxis

Lead end-to-end data strategy, including mining, auto-labeling, and dataset construction to power our ML flywheel

Lead the full post-training stack for VLMs and VLAs, including Continual Pre-training (CPT) on domain-specific driving data, Supervised Fine-Tuning (SFT) for instruction following.

Utilize our large-scale data pipelines and ML infrastructure to research, prototype, and deploy solutions that improve driving behavior

Partner with cross-functional teams to integrate perception signals

MS or PhD in Computer Science or related field

Background in deep learning solutions for VLM and VLA models

Track record in post-training large-scale models, CPT, SFT, RL

Hands-on experience with production ML pipelines, including dataset creation, training frameworks, and metrics

Expertise in Python libraries (PyTorch, NumPy, Pandas, VLLM)

Deep knowledge of cutting-edge computer vision techniques

Publications in top-tier conferences (CVPR, ICCV, RSS, ICRA)

Experience with integrating large language models to various tasks.

Design Evaluation Frameworks: Architect statistical methodologies for safety-critical AI systems to form objective, rigorous conclusions about their performance and reliability.

Conduct Robust Analysis: Deliver validation evidence to support increasingly complex operations and identify potential edge-case failures.

Inform Strategy: Deliver clear, data-driven insights to development teams to guide system improvement, and to executive leadership to inform milestone-level go/no-go decisions.

Define Metrics: Drive alignment across engineering teams on performance metrics and data extraction strategies.

Lead the Lifecycle: Manage all phases of evaluation including prototyping, requirements capture, design, implementation, and validation.

Scale Pipelines: Partner with engineers to build and maintain scalable data processing and simulation pipelines, applying distributed computing to analyze petabytes of driving data.

Hands-on experience with production machine learning pipelines: dataset creation, training frameworks, metrics pipelines

Experience with modern data processing technologies such as Apache Spark, Spark SQL, and Databricks

Experience with designing metrics and delivering actionable insights that drive  business decisions