Ben Cheng
@bencheng
I improve AI training data quality through rigorous annotation, evaluation, and multilingual audio analysis.
AustraliaWhat I'm looking for
At Deakin University, I built and annotated a 1,362-utterance meeting-speech dataset, documenting the rationale behind every labelling decision and reviewing borderline cases. I also audited AI-generated emotion labels across four meeting datasets and LoRA-fine-tuned an open-source LLM to 87.9% classification accuracy, with results published at APSEC 2025.
I'm completing a PhD on multimodal emotion recognition and AI training data quality, combining Mandarin and English speech annotation, prosodic analysis, qualitative coding, and statistical analysis. My six years of live Mandarin radio hosting bring strong multilingual transcription, vocal precision, and sensitivity to accent, register, and emotional tone.
Experience
Work history, roles, and key accomplishments
Conducting research on AI training data quality and multimodal emotion analysis, including building a 1,362-utterance annotated dataset and auditing AI-generated labels. Fine-tuned an LLM using LoRA, achieving 87.9% classification accuracy, and published findings at APSEC 2025.
Education
Degrees, certifications, and relevant coursework
Deakin University
Doctor of Philosophy, Information Technology
2022 -
PhD in Information Technology with a thesis on multi-modal emotion recognition in requirements engineering, focusing on AI training data quality.
Availability
Location
AustraliaAuthorized to work in
Job categories
Interested in hiring Ben?
You can contact Ben and 90k+ other talented remote workers on Himalayas.
Message BenGet matched with your dream remote job
Sign up now and join over 250,000+ remote workers who receive personalized job alerts, curated job matches, and more for free!
