Job Description:
With the rapid advancement of large models and multi-tool invocation capabilities, intelligent assistants must balance intent understanding, dynamic planning, and anomaly handling in complex scenarios. Chain-of-Thought (CoT) and self-reflection mechanisms enhance internal reasoning and self-checking abilities, while automated agents promise intelligent scheduling from instruction parsing to multi-tool closed-loop execution. This achieves a fully closed-loop process from multimodal input to task completion, ultimately delivering more efficient and intelligent service experiences to users.
Join the AutoNavi Maps Intelligent Assistant Agent R&D team, focusing on "intelligent search-recommendation interaction + multi-tool integration."
Design and optimize NLU capabilities based on large language models and reinforcement learning, building automated agent prototypes with chain-of-thought and self-reflection mechanisms.
Conduct algorithm iteration experiments and collect user feedback across core scenarios including navigation, ride-hailing, POI search/recommendation, and local services, participating in iterative improvements.
Contribute to building multi-tool invocation scheduling frameworks, implementing task decomposition, tool interface management, and execution strategy optimization.
Collaborate deeply with cross-functional teams including engineering architecture, data, and product to drive research outcomes into engineering implementation.
Position Requirements:
Master's degree or higher from a nationally recognized university in Computer Science, Artificial Intelligence, Software Engineering, or related fields; PhD preferred.
Minimum 3 years of R&D experience in reinforcement learning, natural language processing, or large-model applications; background in intelligent agent assistant development is advantageous.
Proficient in deep learning frameworks (PyTorch/TensorFlow) with practical expertise in RL algorithms (e.g., PPO, DQN).
Understanding of Chain-of-Thought reasoning and reflective mechanism design; ability to independently build and optimize multi-tool invocation systems.
Strong teamwork and project management skills with cross-departmental communication experience.
Preference given to candidates with publications in top international conferences/journals or leadership in open-source projects (GitHub/GitLab).
Proficient English reading and writing skills to follow cutting-edge AI academic and engineering advancements.