About this role
"Business Risk Integrated Control (BRIC) is a global team protecting ByteDance users, platform health, and the authenticity of community and business ecosystems. We build the risk control algorithms, AI technologies, platforms, and infrastructure behind that work, across products including TikTok, CapCut, Lemon8, and Lark.
Our work spans account and content integrity, and safety across advertising and e-commerce. We apply machine learning across areas such as graph learning, anomaly detection, and multimodal models to understand users, content, behavior, and the relationships between them. The challenge is recognizing genuine activity accurately, so the people using our products never notice the systems working.
You will shape the trust, safety, and sustainable growth of ByteDance's global platforms, applying advanced AI to some of the hardest real-world problems in the field."
Program Intern (Conversion-Oriented): for graduating students who are looking for full-time conversion as the primary objective.
We are looking for talented individuals to join us for an internship. PhD internships at Our Company provide students with the opportunity to actively contribute to our products and research, as well as to the organization's future plans and emerging technologies.
Our dynamic internship experience blends hands-on learning, enriching community-building and professional development events, and collaboration with industry experts.
Applications will be reviewed on a rolling basis, so we encourage you to apply early. Please clearly state your availability in your resume (Start date, End date).
Responsibilities:
- Engine performance, continuously improved by Agents: Build agent workflows that profile the risk engine, identify performance bottlenecks (latency, CPU, memory, cost), propose and verify optimizations, and detect performance regressions. Explore RSI (Recursive Self-Improvement) loops — propose, benchmark, evaluate, iterate — so optimization becomes a sustained, closed-loop process rather than one-off tuning.
- Proactive rule governance by Agents: Design agents that actively audit the rule base across its lifecycle — detecting redundant, conflicting, expired, or low-value rules, analyzing online impact, and generating safe change or decommission proposals with human-in-the-loop approval, rollback, and audit trails.
- Agent platform and evaluation foundations: Contribute to the shared capabilities these agents depend on: tool use and workflow orchestration, evaluation benchmarks and guardrails, observability, and safety controls — and take prototypes to production landing with measurable impact.