Staff Machine Learning Engineer, Content Understanding & Moderation

Written by the T&S Network from Unity's posting.

Multimodal content understanding and moderation for the Unity Ads network, covering creatives and the mobile games they run in. The team turns video, image, playable, audio and text content into model signals used for both ad relevance and detecting content that violates policy, regulation or advertiser and publisher expectations.

What you'd do

• Turn legal and contract requirements into targets a model can be scored on.
• Set standards for quality when labels are sparse, noisy and disputed, and build the tooling to measure them.
• Create shared pieces like content embeddings, game features, eval harnesses and labeling pipelines.
• Set architecture direction across teams, judge designs, and coach senior and mid-level engineers.

What it asks for

• Deep hands-on work with models that combine vision and language, video especially.
• Significant time owning a production trust and safety, moderation, brand safety or policy enforcement system.
• Experience has your teams working where labels are scarce, ambiguous or disputed and violation definitions shift over time.
• Adapting foundation models for production through fine-tuning, distillation and latency and cost work.
• Strong Python and experience running production ML on a live serving path.
• Taking responsibility for production code you review, including code written with AI help.
• Proven ability to lead technical direction across teams without formal authority.

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on Unitynorth_east before you apply. If we got something wrong, report this listing.