Data Scientist (Web3)
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Описание вакансии
TL;DR
Data Scientist (Web3): Designing and developing AI-powered financial trading agents based on Web3 data with an accent on post-training of large-scale language models. Focus on implementing SFT and Reinforcement Learning techniques to enhance trading skills and building evaluation benchmarks.
Location: Must be based in Asia (Work-from-home arrangement)
Company
is a leading global blockchain ecosystem and the world’s largest cryptocurrency exchange by trading volume and registered users.
What you will do
- Design and develop AI-powered financial trading agency skills based on Web3 data and integrate them into practical systems.
- Perform post-training of LLMs using SFT and RL to improve the utilization of internal trading skills.
- Build and maintain benchmark datasets to evaluate AI system performance.
- Iterate workflow prototypes for financial AI agent systems.
- Collaborate with senior engineers on system design, integration, and deployment.
Requirements
- Pursuing or holding a degree in Computer Science, AI, Financial Engineering, Automation, or a related field.
- Hands-on experience with SFT and Reinforcement Learning (RLHF / RLAIF).
- Proficiency with frameworks like Llama Factory, Slime, Verl, or OpenRLHF.
- Experience with PyTorch or TensorFlow and toolkits like Hugging Face Transformers, DeepSpeed, or vLLM.
- Ability to commit 5 days per week for at least 6 months.
- Based in Asia.
Culture & Benefits
- Opportunity to work within the world’s leading blockchain ecosystem.
- Collaboration with world-class talent in a flat, user-centric global organization.
- High degree of autonomy in an innovative, fast-paced environment.
- Competitive salary and company benefits.
- Flexible work-from-home arrangements.
Будьте осторожны: если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого - это мошенники. Обязательно жмите "Пожаловаться" или пишите в поддержку. Подробнее в гайде →