Teams at Tencent

Research and develop omni-modal multimodal large models, including training data, foundational algorithms, pre-training, supervised fine-tuning, reinforcement learning, evaluation, and downstream applications. Analyze performance bottlenecks, develop first-principles solutions, explore next-generation architectures, and advance multimodal understanding and generation. The role also involves optimizing deep learning operators, model training and inference, CPU/GPU acceleration, and distributed systems.

Research and develop omni-modal multimodal large models, including training data, model architectures, pre-training, supervised fine-tuning, reinforcement learning, evaluation, and downstream applications. Analyze performance bottlenecks, develop first-principles solutions, optimize training and inference, and explore next-generation multimodal understanding and generation paradigms.

Coordinates data sourcing, vendor and partner relationships, dataset evaluation, onboarding into AI pre-training pipelines, and data governance tracking. Maintains documentation, dashboards, delivery timelines, quality metrics, licensing records, and compliance logs while facilitating communication among AI research, legal, privacy, security, and external data-provider teams.