Software Engineer - Perception

WatneySan Francisco, CA

About The Position

At Watney, the perception loop is the critical foundation that allows our robots to operate reliably in unstructured environments. This is an opportunity to build the high-throughput data engines, distributed streaming architectures, and deterministic evaluation platforms that power our entire perception lifecycle from raw physical telemetry to deployed edge models. We are looking for someone who wants to design and scale the platform layer that orchestrates our core perception workflows, including automated data mining, multimodal ingestion pipelines, and parallelized regression backbones. The ideal candidate is an experienced backend or data infrastructure engineer eager to build the foundational data engine for autonomous fleets.

Requirements

  • Have experience building high-throughput data platforms, large-scale distributed streaming systems, or infrastructure for autonomous robotics.
  • Have a proven track record designing scalable data pipelines and storage solutions tailored for heavy multimodal media formats (high-frame-rate video, unstructured telemetry).
  • Bring deep experience building closed-loop testing infrastructure, regression frameworks, or automated metric tracking engines for highly complex, non-deterministic software systems.
  • Exhibit a high technical competency in geometric computer vision fundamentals, 3D spatial data handling, or metadata orchestration for machine learning workflows.

Responsibilities

  • Own the Perception Data Engine: Architect and optimize scalable off-robot data pipelines capable of ingesting, parsing, and managing petabyte-scale video and sensor streams for auto-annotation and localization tasks.
  • Build Continuous Evaluation Infrastructure: Design and implement deterministic simulation, regression testing, and evaluation loops to rigorously benchmark perception performance before edge deployment.
  • Scale Annotation & Feature Extraction: Develop high-throughput pipelines that orchestrate multimodal data fusion, automated labeling, and feature extraction to constantly mine high-signal training examples from the field.
  • Drive Quality Metrics & Monitoring: Establish end-to-end telemetry systems and observability frameworks to track model regression, data drift, and perception accuracy across live operational environments.
© 2026 Teal Labs, Inc
Privacy PolicyTerms of Service