Yanfei
Verified
5+ Years in AI Engineer
Contact this Expert
AI Engineer
$-
Saitama, Japan
Prefferd tech
Tech Stack: Fortran, C, Python, TensorFlow, PyTorch, Pandas, NumPy, Matplotlib, BERT, GPT, Vicuna, Deepspeed-chat, Yolov5, Yolov8, Yolox, Yolo-NAS, Transformer (DETR ViT), ANSYS, LMS Amesim, Matlab, Comwave
Tech Skills: ML, NLP, Data Analyst
Expert’s cases:
Led a project on the collection of driving data for ADAS/AD development. (4 team members)
Developed a system capable of collecting data from Lidar, cameras, GPS, and other sensors. This system plays a crucial role in the Data-Driven Development (DDD) pipeline of ADAS/AD and the collected data will be used for AI training, virtual validation, and scenario generation.
This system enabled data measurement with a simple structure, which reduced the cost of dSPACE GmbH’s measurement system by 80%.
Worked on the development of AI image detection models.
Conducted research on image detection AI and successfully developed several AI models capable of detecting custom items. Established a whole process from image collection, labeling, and model training. The cost of outsourcing any model is about $50,000.
Utilized AI models to tag large amounts of data, enabling quick and efficient data retrieval.
Presented research findings at oral conferences.
Utilized AI tools to drive Digital Transformation (DX) within the company.
Investigated the utilization of generative AI, specifically ChatGPT, for the development of a support chatbot.
A local generative AI similar to ChatGPT is being set up. If successful, it could save 50% support time.
Successfully completed a project for a sports manufacturer using the Japanese version of BERT.
Conducted sentiment analysis with and without user benefits using the Japanese version of BERT.
Made modification to improve the verification accuracy rate of the model.
Developed a Chinese version of the BERT model for a project with an electric machine manufacturer.
Conducted survey, implementations and sentiment analysis reasoning confirmation using the Chinese version of BERT
Analyzed reviews on a Chinese e-commerce site for a certain electrical equipment manufacturer