AI Model Reinforcement Russian Speaking Graduate (Safety Model Operations) - 2027 Start

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

Builds and maintains machine learning safety models and operational processes that support TikTok's Trust & Safety systems, with a focus on training and evaluation data for large language models. The team works on automated detection of harmful content across regions, and this graduate role is tied to the Russian market.

• AI misuse and jailbreaks

What you'd do

• Help define and keep updating the standards, judgment rules, and acceptance bars used to grade model training and evaluation data.
• Review data outputs against quality standards, flag issues, and close the loop on corrections.
• Dig into murky and borderline cases, then convert the outcomes into rules the team can reuse.
• Track quality performance, find root causes, and drive fixes in production processes.
• Coordinate with production, product, algorithm, and strategy teams to align the standards, production, acceptance, and feedback cycle.

What it asks for

• A bachelor's degree, with majors in legal studies, philosophy, economics, media, languages, or social sciences favored.
• Prior work on data quality, moderation, compliance, content ops, AI data, or labeling, through internship or project (preferred).
• Strong English across all skills, able to follow English video and text precisely; IELTS 7.5, TOEFL 105, or equivalent.
• Russian proficiency, required for daily liaison with the local Russian market team.
• Familiarity with Excel and common data-processing tools.
• Hands-on use of major foundation-model tools like ChatGPT, Claude, or Gemini, plus a working grasp of their limits (preferred).

Worth knowing

• Graduate role with a required onboarding date by the end of the year.

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