- 経験
- 3年以上
- 給料
- USD 100 – USD 120 / hour
- 求人情報
- 1
- 投稿済み
- 2週間前
- 作業モード
- 在宅勤務
- 教育
- Degree from a top-100 university or equivalent experience
- 再開する
- 応募必須
仕事内容
About the Role
Mercor, a company that connects top-tier creative and technical talent with leading AI research institutions, is offering a contract opportunity for a LLM Research Scientist focusing on both pre-training and post-training phases. The role is fully remote and part-time, offering competitive hourly compensation between $100 and $120.
Key Responsibilities
- Develop and train transformer-based language models starting from initial phases as well as fine-tune pre-existing open-weight models.
- Enhance model performance, especially under limitations related to data availability and computational resources.
- Create training datasets by compiling large-scale raw web data sources.
- Design and maintain post-training workflows and pipelines.
- Identify and troubleshoot issues encountered during training processes to improve model quality and effectiveness.
- Operate autonomously and asynchronously to achieve defined project objectives.
Qualifications
Essential Requirements:
- At least three years of professional experience in machine learning research.
- Proficiency with machine learning frameworks such as PyTorch, JAX, TensorFlow, or equivalents.
- An academic degree from a top 100 ranked university, prior employment at a major AI tech company (e.g., FAANG), or similar robust research background.
Preferred Skills and Experience:
- Familiarity with scaling laws, curriculum learning techniques, and evaluation methodologies for large language models.
- Understanding of reinforcement learning applications within language modeling.
- Knowledge about model alignment practices and AI safety considerations.
Application and Support
- The application process includes uploading a resume, completing an AI-based interview related to the submitted resume, and submitting the application form; this takes approximately 20 to 30 minutes.
- For questions or assistance during the application, candidates can reach out to support@mercor.com.
- Applications are reviewed regularly; applicants should complete all steps, including the AI interview, to be considered.