ML Engineer, Trust & Safety [Hapiko]

Spin Master Ltd•New York, NY

9d•CA$136,700 - CA$182,300•Remote

About The Position

Hapiko, a Spin Master company, makes Stickerbox, a voice-activated sticker printer for children. The ML Engineer, Trust & Safety will ensure that stickers printed by the product are safe as the product scales globally. This is an engineering role focused on owning and improving the moderation pipeline, evaluation systems, and supporting tools, with a focus on maintaining strong protections while allowing harmless content through. The role owns end-to-end content moderation, from a child's spoken request through the image that is printed, including content involving copyrighted and licensed characters. It also covers safety evaluations, classifier improvements, red teaming, review tooling, incident response, and the technical aspects of compliance obligations such as data retention and deletion. The engineer will partner with ML and Backend / Platform engineers to build safety into new models and features before release.

Requirements

Applied safety engineering experience on a product with real users.
Experience building evaluations or classifiers for harmful content, rather than only configuring a vendor API.
Strong Python software engineering skills, including the ability to ship production code in a moderation path that must be fast and fail-safe.
Comfort with precision/recall tradeoffs and the judgment to set thresholds when both types of error have a cost.
Good judgment under pressure.

Responsibilities

Own and maintain end-to-end content moderation, from what a child says to the image that prints, including copyrighted and licensed characters.
Maintain and expand safety evaluation suites, and use them to assess whether model, prompt, or threshold changes are ready to ship.
Continuously improve classifiers and determine where in-house and third-party models are best suited.
Tune moderation to reduce over-blocking without weakening protection.
Conduct ongoing red teaming against models and the moderation pipeline.
Maintain review tooling for flagged content and parent reports.
Own the incident response runbook for safety issues.
Maintain the technical aspects of compliance obligations, including data retention and deletion.
Partner with ML and Backend / Platform engineers to incorporate safety into new models and features before they ship.