Job Description

Midea is a leading Global Fortune 500 high technology company. Committed to bring innovation to life, Midea Group aims to develop solutions for our international customers that not only meet the requirements of the present, but also the challenges of tomorrow and the day after tomorrow. Midea develops, produces, and sells innovative products in five competence areas: Smart Home, Industrial Technologies, Building Technologies, Robotics & Automation, and Digital Innovation. We operate in more than 195 countries with over 150,000 employees. Midea embraces talents of characters of aiming high, customer first, transformation and innovation, inclusion and partnership to join the mission together to integrate with the world, and to inspire your future!

AI empowers Midea by driving innovation across its diversified applications, enhancing efficiency, intelligence, and user experience. From industrial robotics and automation, smart home appliances, intelligent manufacturing, health care sector and energy solutions, AI enables Midea to optimize operations, improve product performance, and deliver personalized experiences. Through cutting-edge machine learning, computer vision, and IoT integration, Midea leverages AI to stay at the forefront of technological advancement, creating a smarter and more connected world.

We are hiring Innovative AI Engineers who will drive Midea AI journey to next level.

Job Description

1. Efficient Implementation of Reinforcement Learning Training and Development of a Unified and High-Performance Reinforcement Learning Training Framework.

2.Develop reinforcement learning algorithms for large language models to enhance training efficiency during the reinforcement learning phase and improve reasoning capabilities in natural sciences such as mathematics and coding.

3. Develop reward and evaluation models, including fine-grained process supervision and reward modeling, covering tasks such as complex reasoning and instruction following.

4. Participate in Scaling Law research during post-training and inference stages, including reward model training, reinforcement learning training, and inference phase Scaling Law analysis.

Job Requirements

1. Master’s degree or PHD in Computer Science or a related field.

2. Research experience in large language models, with hands-on training experience in post-training, and familiarity with reward model modeling and mainstream reinforcement learning algorithms such as PPO, REINFORCE, and RLOO.

3. Strong algorithm engineering skills, proficiency in Python programming, and experience with the PyTorch deep learning framework. Familiarity with mainstream distributed training frameworks such as DeepSpeed and Megatron.

4. Strong analytical and problem-solving abilities, excellent engineering practices, and the ability to think independently and solve real-world problems.

5. Strong teamwork and communication skills, with the ability to collaborate closely with engineering, business, product, and technical teams.

Preferred Qualifications:

1. Research or practical experience in large language models and machine learning, with high-quality publications in top international conferences/journals.

2. Research or practical experience in big data processing, large-scale distributed computing, and distributed training.

Midea America Corp. is an equal opportunity employer. We do not discriminate based upon race, religion, color, national origin, gender (including pregnancy, childbirth, or related medical conditions), sexual orientation, gender identity, gender expression, age, status as a protected veteran, status as an individual with a disability, or other applicable legally protected characteristics.

Job Tags

Similar Jobs