Responsibilities:

• Design, implement, and evaluate training      pipelines for large generative AI models, encompassing multiple      stages of post-training.
•   
• Data: Design,      implement, and evaluate data augmentation pipelines to increase the      diversity and robustness of training datasets, improving model      performance, particularly in low-data regimes.
•   
• Model evaluations: Develop and implement model evaluation      pipeline for LLMs
•   
• Supervised Fine-tuning (SFT): Developing and executing SFT strategies for      specific tasks.
•   
• RLHF:      Developing and leveraging RLHF algorithms in model training such as DPO      and KTO
•   
• Reinforcement Learning (RL): Exploring RL training strategies, sampling,      reward function design and etc. to apply large scale RL to model training
•   
• Quantization: Implement and evaluate model quantization techniques to      reduce model size and accelerate inference speed, balancing precision loss      with performance gains for deployment across diverse hardware platforms.
•   
• Low-Rank Adaptation (LoRA): Utilizing techniques for efficient      fine-tuning of large language models, balancing performance and resource      constraints, and tailoring model performance      for downstream tasks well.
•   
• Experiment with various training techniques,      hyperparameters, and model architectures to optimize performance      and efficiency.
•   
• Develop and maintain data pipelines for      processing and preparing training data.
•   
• Stay up-to-date with the latest      advancements in large language models, training techniques, and related      technologies.
•   
• Collaborate with other engineers and researchers      to design, implement, and deploy AI-powered products.
•   
• Contribute to the development of internal tools      and infrastructure for model training and evaluation. 

 Qualifications:

• Master's degree in Computer Science, Machine      Learning, or a related field and 2+ years of relevant work      experience or 4+ years of relevant work experience.
•   
• Strong programming skills in Python and      experience with deep learning frameworks like PyTorch.
•   
• Solid understanding of machine learning      principles, including supervised learning, unsupervised learning, and      reinforcement learning.
•   
• Proven experience in designing and conducting      experiments, analyzing data, and drawing meaningful conclusions.
•   
• Familiarity with large language models,      transformer architectures, and related concepts.
•   
• Experience with data processing tools and      techniques
•   
• Excellent communication, collaboration, and      problem-solving skills.