Software Engineer, Labelling, Data & Automation

Waabi Waabi · Robotics · Toronto, ON +3 · Remote · Software Engineering

Software Engineer on the Labelling and Data Automation team responsible for building pipelines, tools, and workflows to accelerate the development of AI-first self-driving technology. This role is critical in the training pipeline, focusing on data extraction, labeling, and integration for ML scientists and engineers. It also involves deploying open-set/embedding models for search and curation.

What you'd actually do

  1. Design and implement tools, pipelines, and metrics to accelerate the development of our AI-first autonomy system and generative AI simulator.
  2. Own the process, criteria, and tooling for efficiently finding interesting and relevant data across the petabytes of real world data that Waabi has collected
  3. Build high reliability systems for extracting and labelling the interesting data with various vendors and integrate it back into our system
  4. Work with both internal and third party stakeholders to define taxonomy, validation rules and success criteria for our labelling projects
  5. Design and manage the end-to-end deployment of data solutions to deliver high quality labelled data for various ML teams to use in experiments and model improvement

Skills

Required

  • Python programming
  • strong software engineering fundamentals
  • high quality, well-structured, and well-tested code
  • data pipelines for large-scale processing and analysis
  • cloud job orchestration, monitoring, and instrumentation best-practices

Nice to have

  • ML pipelines, including dataset curation, labelling, training and evaluation
  • self-driving technology or related fields
  • linear algebra (projections, transforms) and 3D geometry
  • MapReduce frameworks (Apache Hadoop/Spark) or orchestration frameworks (Apache Airflow/Apache Beam/Google Dataflow/AWS Step Functions)
  • front end development
  • open-set / embedding models and deploying them in a production setting
  • infra as code (Terraform, CloudFormation, etc)

What the JD emphasized

  • AI-first self-driving technology
  • petabytes of real world data
  • open-set / embedding models
  • high quality labelled data

Other signals

  • building practical, scalable tools and automation for extracting interesting and relevant data
  • creating high-quality ground truth labels
  • providing this data to our expert machine learning scientists and engineers
  • deploy open-set / embedding models to our production environment