Algorithm Researcher — FULL_TIME
Key Responsibilities
- Take ownership of the full-stack algorithm R&D for post-training of Vision-Language Models.
- Explore Supervised Fine-Tuning and reinforcement learning alignment algorithms like RLHF, PPO, and GRPO.
- Optimize low-level visual perception, fine-grained understanding, and spatial awareness.
- Explore Visual Agent solutions including thinking with images to improve physical world execution.
- Reduce hallucinations in multimodal large models by investigating root causes and building detection pipelines.
- Implement refined data cleaning strategies and lead R&D for reinforcement learning-based hallucination removal.
- Align and optimize models for real-world user scenarios to strengthen logical reasoning and instruction following.
- Build an online badcase monitoring and data flywheel feedback loop to refine interaction experiences.
Job Requirements
- You must possess a Master's degree or higher in Computer Science, Software Engineering, Artificial Intelligence, or a related field.
- You need a deep understanding of Transformer architecture and extensive hands-on experience in post-training domains like LLM alignment or RLHF.
- You should have solid engineering skills in Python and proficiency with deep learning frameworks such as PyTorch or TensorFlow.
- You are required to have practical experience with distributed training and large-scale frameworks like Megatron-LM, DeepSpeed, or vLLM.
- You must have experience in fine-tuning models at the tens-of-billions to hundreds-of-billions parameter scale.
- You should demonstrate strong research acumen through publications at top-tier conferences or high-impact open-source contributions.
Job Details
