Job Details

About the Company: A high-tech company specializing in high-end equipment & instruments. It is recruiting overseas high-level talent in areas such as Speech Recognition Algorithm Engineer.

Target Candidates: Overseas PhDs or senior algorithm experts in AI, large language models, multimodal algorithms, and machine learning engineering

Responsibilities

    •  Responsible for research on voice interaction scenarios, including design and development of speech recognition algorithms, false-wake suppression methods, and custom wake-word algorithms.

    •  Responsible for engineering-project development and maintenance, including data processing, model training, network-structure optimization, and partial feature tooling.

    •  Follow the latest industry technologies, solve existing problems with new methods, and achieve rapid deployment of new technologies.

    •  Plan the technology roadmap for LLMs, multimodality, computer vision, recommendation/prediction, and AI engineering, and drive the productization of algorithms.

Requirements

    •  PhD in Mathematics, Computer Science, Automation, Signal Processing, or related fields; project R&D experience in speech recognition or custom wake-word detection is preferred; fresh PhD graduates with research backgrounds in deep learning or speech signal processing may be considered with relaxed requirements.

    •  Has basic engineering ability and is proficient in one or more programming languages such as Python or C.

    •  Familiar with common open-source speech recognition frameworks such as SpeechBrain, PaddleSpeech, and WeNet.

    •  Familiar with common open-source speech recognition frameworks such as SpeechBrain, PaddleSpeech, and WeNet.

Highlights & Support: This recruitment covers approximately 2 positions with compensation of RMB 500K–600K/year. Specific position level, compensation, and support terms can be further negotiated based on the candidate's academic background, project experience, commercialization achievements, and interview results.