Haotian Cui Email and Phone Number

- Organize schedules and resources for data collection of LLM;- Research on data sources and means of acquirement for LLM;- Build relevant datasets for model training or evaluation

- Research: Studied rules of target speech phenomena, e.g., acoustic features, variations and contexts in collaboration with algorithm engineers as a standard, based on which designed principles and instructions for recording and annotation- Teamwork: Participate in speaker selection; afterwards monitored the recording process to ensure the speech data are up to the standard- Organizational skills: Responsible for checking and accepting audio data (around 8,000 recorded sentences per project) from speakers and coordinated outsourced annotation team- Technical: Ran labelled data according to established processing procedures on company’s server with Linux Commands under time pressure- QA: Rated synthesized speech and analysed reasons of bad cases with algorithm engineers

- Inspected labelled data on an intention recognition projects to identify what the negative feedback concerned, where I suggested the manager to change original categories from three to five based on labelled data and thus increased the accuracy as well as made the model more suitable for the project;- Annotated dialogues on an automatic customer service chatbot task which is already in use in an on-line education platform.

- Provided manually corrected sentence pairs to enhance AI auto-correction accuracy;- Labelled error categories for an error-type classification task.

- Assisted other teachers’ classes and jobs;- Organized English-speaking practice sessions with small groups and extra-curriculum workshops;- Participated in faculty’s discussions about teaching plans;- Taught academic English with regular teachers.