Machine Learning Operations Manager (Safety Model Operations) - AI Data Service & Operations

Responsibilities

Our Company will be prioritizing applicants who have a current right to work in Singapore, and do not require Our Company's sponsorship of a visa.

About the Team
The AI Data Service and Operations (ADSO) team is responsible for providing safety and non-safety data annotation services and search operation services for all of the company's international products, which can also help international products build their own data ecological security.

The Safety Model Operations (SMO) team within ADSO is responsible for building, optimizing, and maintaining machine learning (ML) models and operational processes that support TikTok's Trust & Safety systems. We ensure that automated safety models perform effectively in identifying harmful content, mitigating risks, and maintaining a safe user environment across regions.

What You Will Do
- Own and optimize the end-to-end Machine Learning Operations (MLOps) lifecycle for safety models, including data ingestion, feature engineering, model training, evaluation, deployment, and monitoring.
- Design, implement, and continuously improve automated ML pipelines (training, inference, and evaluation) to increase efficiency, scalability, and reliability.
- Collaborate with ML Engineers, Data Scientists, and other relevant cross functional teams to identify model performance gaps, data drift, and production issues, then drive initiatives to resolve them.
- Explore and integrate the latest MLOps tools, frameworks, and LLMOps best practices to streamline experimentation, model versioning, CI/CD for ML, and automated retraining.
- Establish and maintain robust model monitoring, observability, and alerting systems to detect performance degradation, bias, and data quality issues in production.
- Lead initiatives to improve training data quality and coverage by developing advanced data pipelines, synthetic data strategies, and active learning approaches.
- Take end-to-end ownership of model operations - from dataset curation and standard setting to production deployment and post-launch performance tracking.
- Stay current with cutting-edge research and industry trends in MLOps, LLMOps, and scalable model serving.

Qualifications

Minimum Qualifications
- Bachelor's or Master's degree in Computer Science, Data Science, Machine Learning, Engineering, or a related quantitative field.
- 3+ years of experience in MLOps, LLMOps, or production ML systems.
- Proven experience to lead and drive end to end projects.
- Strong understanding of machine learning fundamentals, including LLMs, transformers, and modern deep learning architectures.
- Hands-on experience with MLOps practices and tools (e.g., MLflow, Kubeflow, Airflow, Weights & Biases, or similar).
- Proficiency in Python and experience with ML frameworks such as PyTorch or TensorFlow.
- Solid experience in building and maintaining data and ML pipelines, cloud platforms (AWS, GCP, or Azure), containerization (Docker), and orchestration (Kubernetes).
- Demonstrated ability to manage cross-functional projects, with excellent communication skills (technical and non-technical) and a strong sense of ownership.

Preferred Qualifications
- Experience in Trust & Safety, content moderation, or working with large-scale moderation models.
- Strong knowledge of CI/CD for machine learning, model serving (e.g., Triton, TorchServe, SageMaker), feature stores, and monitoring tools (e.g., Prometheus, Grafana, Evidently).
- Experience with big data technologies (Spark, Hadoop, Flink) and large-scale distributed training.
- Familiarity with regulatory compliance and ethical AI practices in content safety.