About this role
Developing the automation layer for dental laboratories, this AI/ML Engineer establishes precise accuracy reporting, builds correction flywheels, standardizes squad deployments, solves multi-tenant model behavior, and scales production inference. The team tackles the core challenge of automating a complex manufacturing pipeline involving 3D scans and messy handwritten prescriptions across thousands of global customers.
EviSmart exists because current automation falls short of running unattended, making model reliability the company's highest-leverage technical problem. An ideal candidate possesses deep platform experience, evaluation rigor, and the ability to align cross-functional engineers without formal reporting authority, thriving in a hands-on, in-office environment.
What you will do
- Establish per-model, per-customer, and per-category accuracy reporting that Product and Sales can actionable rely on.
- Capture customer quality-control corrections, transform them into structured training data, and integrate human review loops.
- Standardize deployment paths, evaluation harnesses, and model registries across four distinct engineering squads.
- Address multi-tenant model behavior to handle diverse customer product codes and preferences over a shared model base.
- Migrate inference workloads from on-premise infrastructure to the cloud while optimizing inference costs and concurrency.
- Partner closely with the CTO, Product, and operations teams to set overarching technical direction for AI across the company.
What you need
- Significant experience building and operating production machine learning platforms including model registries, versioned datasets, and monitoring.
- Proven depth in document and information extraction from unstructured, inconsistent text such as handwritten prescriptions.
- Demonstrated experience designing systems where user corrections directly improve subsequent models.
- Genuine evaluation rigor with the ability to define, defend, and ship against strict accuracy metrics.
- Strong technical judgment regarding when to apply deterministic rules instead of machine learning.
- Demonstrated ability to align engineers across squads and time zones without formal reporting authority.
Nice to have
- Multi-tenant or per-customer model behaviour at scale
- 3D or geometry-based ML for scan quality and restoration design
- Experience with healthcare or regulated data, patient privacy, and data residency
- Inference cost management and on-premise to cloud migration
Skills and tech stack
Questions candidates ask
What is the remote work policy for this position?
This is a full-time in-office role at the Vancouver office, five days a week. Remote or hybrid arrangements are not offered and are not negotiable at the offer stage.
What does success look like in the first 90 days?
You must be able to tell the team, with evidence, where the models actually fail and for which customers. Additionally, the squads must agree on a single evaluation standard.
Is a background in dental or medical devices required?
A dental or medical-device background is not required. The workflow is learnable in an afternoon, and the team values platform depth over domain history.