3+ years as a software engineer managing production services in the cloud, ideally Azure.
Hands-on experience deploying machine learning models using frameworks like Triton or SageMaker.
Proficient in Docker, Kubernetes, and CI/CD practices with versioned releases.
Experience with monitoring tools such as Prometheus or Grafana and handling real production incidents.
Strong Python skills and familiarity with PyTorch, especially in computer vision contexts.
Ability to document processes clearly for cross-team collaboration.
Effective communication skills to explain technical issues to non-engineers.
Responsibilities
Scale AI products on cloud GPU infrastructure to meet demand and control costs.
Build and manage the deployment pipeline from model development to production.
Set up monitoring and alerting systems for uptime and performance metrics.
Coordinate with CAD design and QC teams to address issues effectively.
Provide reliable training infrastructure and reproducible pipelines for AI engineers.
Benefits
Onsite work in a vibrant Vancouver office.
Opportunity to work closely with AI and design teams.
Shared on-call responsibilities for production support.
Individual contributor role with potential for growth.
Full Job Description
On-site | Vancouver Office - 564 Beatty St
You9ll take our AI products from working models to reliable products running at scale in the cloud, and own keeping them running.
Our AI models are ready for wide use. This role builds the engineering that lets them run fast, reliably and cost-effectively at scale. What you9ll do
Scale our AI products in the cloud. Run our AI models on cloud GPU infrastructure that scales with demand and keeps costs under control.
Own deployment end to end. Build the pipeline that takes a model from the AI team to production: packaging, versioning, staged releases and fast rollback when something goes wrong.
Keep it running. Set up monitoring and alerting for uptime, speed and model performance, so we hear about a problem before a lab does. Respond to incidents, fix the cause and write down what changed.
Work with the people who use what you ship. Coordinate releases with the CAD design and QC teams, and help triage whether an issue is a model, pipeline or infrastructure problem.
Make the AI team faster. Give AI engineers reliable training infrastructure, reproducible pipelines and versioned data and models, so they spend their time on models rather than servers.
What you bring
To scale our products in the cloud: 3+ years as a software engineer running production services in the cloud (ideally Azure), where you owned uptime, not just the code. You9ve scaled a service as load grew and kept an eye on what it cost.
To deploy models: hands-on experience putting machine learning models into production with a serving framework (such as Triton, TorchServe, Ray Serve or SageMaker), ideally on GPUs.
To own deployment: Docker and Kubernetes, infrastructure as code (such as Terraform), and CI/CD with versioned releases and rollback.
To keep it running: monitoring and alerting tools (such as Prometheus, Grafana or Datadog), and real production incidents you can walk us through: how you found out, what you did and what you changed afterwards.
To work alongside the AI team: strong Python, and comfort reading PyTorch code. You understand computer vision basics, such as what a classification or segmentation model takes in and puts out, well enough to tell a model problem from an infrastructure one.
To share coverage across time zones: you write things down. Your runbooks and handoff notes let someone else fix your system without calling you.
To work with our design and QC teams: you explain technical issues clearly to non-engineers and enjoy working closely with the people who use what you build.
Nice to have
3D data such as meshes, point clouds or scans, or medical imaging.
Making models faster and cheaper to run (ONNX, TensorRT, batching, GPU autoscaling).
Data and model versioning tools (such as MLflow or DVC) and model registries.
Event-driven or queue-based pipelines that process uploads as they arrive.
Running software in a regulated or healthcare setting.
Working day to day with a team in another time zone.
How this role works
The shared 4How we work4 section in our engineering JDs applies here too. A few points matter most for this role:
Individual contributor. You9ll start as an IC with no direct reports, working alongside the AI team and the teams using your systems.
Onsite, five days a week in our fabulous Vancouver office at 564 Beatty St.
Shared on-call. You9ll cover the Vancouver side of production support. Expect some after-hours responsibility when something breaks.