Experience

Jun 2023 - Jan 2024 Machine Learning Engineer at Sam.coach

– Leading two NLP services including summary and follow-up letter generation used in the sales field, designing and building services to generate an insightful summary and follow-
up letter for assisting sales get more meaningful meeting reports.
– Applying multi-agentic frameworks to make services more flexible and realistic, and leveraging prompt engineering and few-shot learning to generate insightful reports for sales
to analyse clients’ intentions easily.
– Deploying services via server less computing on Azure to save DevOps costs.
– Build a proof-of-concept environment via Streamlit for potential clients to test products to gain a more positive product reputation.
– As a mentor, share working experience and professional skills with an intern to promote work efficiency.

June 2019 - March 2023 Machine Learning Engineer at Emotibot

– Leading two text-to-speech (TTS) services including batch and streaming services used in the financial field, also supporting the automatic speech recognition (ASR) team in
designing and building a predictive model to achieve better user experience in reading comprehension.
– Designed and built and maintained the whole Mandarin & code-switch TTS system into docker which is deployed to 5-10 companies per year.
– Developed a teacher-student training procedure in a TTS system with 0% human labels and saved 20% of development cost.
– Using the state-of-the-art model in prosody, polyphone, and punctuation prediction in Mandarin to increase accuracy in frontend of TTS for generating natural, fluent voice, and
optimising user experience.
– Applied voice adaption in TTS with a small dataset for personalised voice clones to be used in various situations, saving 80% cost and reducing 80% time in the voice clone
procedure.
– Applied ONNX/jit instead of TensorFlow/Pytorch to speed up 1.2-1.5 inference times in TTS system for improving user experience in response time.
– Applied git to version control and unit-test/stress-test before releasing the new version to ensure the service robustness.
– Experience in MLOps such as orchestration, CI/CD pipeline, and Docker/Kubernetes and cloud platforms, such as AWS.