Software Engineer, Ads Tooling

The ads system at TikTok operates on a massive scale and serves millions of advertisers, clients and influencers across the world. The quality of the ads system depends heavily on the underlying infrastructure that handles massive data. The USDS Ads team empowers TikTok's global monetization by efficiently delivering ads and content on TikTok.

Tooling is what makes changing that system fast — and safe. Ads correctness cannot be asserted by comparing an expected string to an actual one: ranking outputs are probabilistic, behavior is driven as much by configuration as by code, and a silent regression in a prediction pipeline costs revenue long before it triggers an alert. Off-the-shelf test frameworks don't cover this, so we build our own. You will build the platforms and tooling that let every engineer on the team validate their changes — automated validation across ads pipelines, and continuous system-health monitoring that protects business metrics. Just as importantly, you'll go looking for the manual, repetitive, and error-prone workflows our engineers live with today, and replace them with automation.

You'll join a team that invests in your growth — with mentorship from senior engineers, ownership of real surfaces from early on, and a clear path toward technical leadership.

Responsibilities

• Build and own the internal tooling and platforms that power ads system validation — test frameworks, environment and data provisioning, assertion libraries, and self-service harnesses that let engineers validate changes without hand-rolling infrastructure each time.
• Identify inefficient engineering workflows and automate them away. Audit how the team actually works today — manual verification steps, repeated debugging rituals, slow feedback loops, high-toil release procedures — quantify the cost, and build the tooling that eliminates it.
• Build quality infrastructure for ads systems, including training/serving consistency checks, prediction-service invariant and fallback validation.
• Build automated system-health monitoring that surfaces regressions in business and quality metrics, routes findings to the owning engineer with enough context to act, and shortens time-to-detection.
• Integrate tooling into CI/CD as release gates, so validation runs by default rather than by remembering to run it.
• Treat internal engineers as your users — measure adoption, gather feedback, and iterate on developer experience rather than shipping tools and walking away.
• Bring in new ideas and tooling, including applying LLM and agent-based approaches to test generation, failure triage, and coverage analysis.

Minimum Qualifications

• Bachelor's degree or above, major in computer science or related technical disciplines.
• 3+ years of hands-on experience in software engineering, developer tooling, quality engineering, or test infrastructure for complex and scalable distributed systems.
• Track record of building tooling or frameworks used by other engineers — not only authoring tests within an existing framework.
• Hands-on authorship of CI/CD pipelines including release-gating logic.
• Demonstrated instinct for spotting repetitive or manual engineering work and replacing it with automation, with measurable results.

Preferred Qualifications

• Experience in the Ads domain — auction mechanics, bidding, ranking, attribution, or billing.
• Experience building internal developer platforms, engineering-efficiency tooling, or test infrastructure adopted across a team or org.
• Experience testing or validating large-scale ranking and machine learning infrastructure, including model evaluation, A/B experiment correctness, or online/offline consistency.
• Applied LLM/agent tooling for engineering workflows, including structured-output schema design.