Product Manager - LLM Training

ByteDance; Singapore; On-site; SGD 60,000 - 80,000; Full time

Job summary

A leading tech company in Singapore seeks a product manager for Big Language Models. You will analyze requirements, collaborate with teams for integration, and ensure data quality. Ideal candidates have over three years' experience in algorithm-driven products, a solid understanding of alignment strategies, and a passion for LLMs and human behavior. Fluency in English is essential.

Qualifications

• 3+ years experience as product manager or algorithm engineer for algorithm-driven products.
• Solid understanding of alignment approaches like RLHF.
• Strong interest in LLMs and human behavior.

Responsibilities

• Analyze requirements and optimize strategy for Big Language Model.
• Collaborate with business teams on LLM integration.
• Monitor data acquisition costs and effectiveness.

Skills

Collaboration with business and technical teams

Data quality management

Understanding of Supervised Fine-Tuning (SFT)

Proficiency in English

Job description

Responsibilities

Responsible for requirement analysis and strategy optimization of Big Language Model, responsible for data quality and model effect, and working with multiple teams such as evaluation and R & D to improve model effect.

• Collaborate with business teams to integrate ByteDance self-developed LLM into new and existing product offerings and services; identify room for improvement of Global Seed LLM from studies of user behaviors and feedback.
• Collaborate with technical teams to precisely define the data requirements necessary for aligning Large Language Models (LLMs) across global markets.
• Develop methodologies for data acquisition and production, while also monitoring costs and assessing the effectiveness and efficiency of these processes. Ensure data quality is maintained at high standards and continually enhances processes based on both quantitative and qualitative feedback.
• Evaluate the impact of data production tools on the effectiveness and quality of data production. Work closely with technical teams to enhance both the quality and effectiveness of these tools.

Qualifications

Minimum Qualifications

• Solid understanding of alignment and related approaches, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF)
• Minimum of three years of experience as a product manager for algorithm-driven products or features, or as an algorithm engineer
• Proficiency in English (spoken and written).

Preferred Qualifications

• A strong interest in both LLMs and human behavior/experience/wellbeing. Must be an enthusiastic learner. This position is not suitable for individuals who find reading papers and books to be tedious.
• Collaborative team player who knows how to assertively advocate for their ideas without being overbearing. Skilled in maintaining attention to detail while staying focused on prioritized goals and the bigger picture.