AI Algorithm Engineer

Job Responsibilities

1. Build Advanced Memory Architecture: Deeply involved in the design and implementation of the EverMemOS four-layer architecture (Memory Layer, Retrieval Layer, Decision Layer, User Layer). You will have the opportunity to optimize every aspect from raw data to the final response.

2. Develop Unified Memory Models: Participate in the design and development of large model-related algorithms, especially language model architecture improvements and parameterized memory model implementation, tackling the integration of dynamic memory, static memory, and parameterized memory models.

3. Design Human-like Decision and Retrieval Strategies: Address cutting-edge topics such as “Multi-Step Retrieval Thinking.” You will imbue the system’s “Decision Layer” with human-like reasoning and develop hybrid retrieval algorithms that integrate keywords, vectors, knowledge graphs, and advanced RAG technologies.

4. Create Industry-Leading Evaluation Benchmarks: Participate in building and improving our EverMemBench, proposing innovative diagnostic frameworks that decouple long-context tasks into retrieval and generation stages for evaluation, thereby promoting the establishment of industry standards.

Job Requirements

1. Bachelor’s degree or above in Computer Science, Artificial Intelligence, or a related field.

2. Proficient in Transformer models, familiar with model architecture design, and has experience in algorithm model design and evaluation.

3. Strong engineering implementation skills, able to handle debugging and performance optimization in the process of algorithm implementation, and familiar with deep learning training frameworks and scaffolding.

4. Eager to solve problems in the AI field, pursue the user value of AI technology, and not focus solely on paper publication; possesses good teamwork and communication skills, is proactive, and has strong analytical and problem-solving abilities.

5. Excellent coding skills, solid foundation in data structures and basic algorithms, and proficient in Python.

6. Excellent learning ability, able to read English documents fluently, and quickly learn and apply new technologies.

Preferred Qualifications

1. Prioritization will be given to candidates who have led influential projects or published impactful papers in the field of model design.

2. In-depth experience in long-term memory, linear/sparse model design, RAG, knowledge graphs, or reinforcement learning is preferred.