Machine Learning Engineer (LLM/VLM) - Trust and Safety (Australia)

We are looking for people like you with solid experience in designing and deploying state-of-the-art models in the combination of NLP and CV related areas. This position will work with a team of excellent research scientists and machine learning engineers who can take initiatives, design and develop advanced machine learning solutions, and deploy them directly to TikTok's global platform.

• Build industry-leading content safety systems for TikTok;
• Develop highly-scalable classifiers, tools, models and algorithms leveraging cutting-edge machine learning, computer vision and data mining technologies;
• Understand product objectives and improve trust and safety strategy and model's performance;
• Work with cross-functional teams to protect TikTok globally.
• Collaborate with data analysts to understand and find data patterns

• 3+ years experience in one or more of the areas: machine learning, deep learning, computer vision, NLP, Audio, content understanding.
• Skilled in pre-training foundation technologies, including efficient training and pre-training as a service, downstream applications of NLP/CV/Video are preferred
• Proficient in deep learning frameworks such as PyTorch and TensorFlow, and programming languages such as python or java.
• Good programming skills in Python and similar languages, and a deep understanding of data structures and algorithms
• Good communication and teamwork skills, be passionate about learning new techniques and taking on challenging problems

• Published papers in the top AI conferences or journals is a plus, including KDD, IJCAI, WWW, WSDM, ICML, NeurIPS, CVPR, ECCV, ICCV, ACL, etc.;
• Experience in trust and safety area is a plus
• 3+ years of experience participating in large-scale online machine learning platforms are preferred

Building a world where people can safely discover, create and connect. The Trust & Safety (T&S) team at TikTok helps ensure that our global online community is safe and empowered to create and enjoy content across all of our applications. We have invested heavily in human and machine-based moderation to remove harmful content quickly and often before it reaches our general community.