• Handled the Honda Titan dataset comprising 257GB of pre-labeled images from car front cameras, verifying the accuracy of labels and ensuring data quality• Developed and implemented Python scripts to automate the conversion of raw datasets into a readable format for the labeling tool and subsequently converting them into the desired dataset format for training, greatly enhancing workflow efficiency • Researching, cleaning, formatting and splitting BDD100K driving video… Show more • Handled the Honda Titan dataset comprising 257GB of pre-labeled images from car front cameras, verifying the accuracy of labels and ensuring data quality• Developed and implemented Python scripts to automate the conversion of raw datasets into a readable format for the labeling tool and subsequently converting them into the desired dataset format for training, greatly enhancing workflow efficiency • Researching, cleaning, formatting and splitting BDD100K driving video dataset, Teledyne FLIR driving video dataset, into the desirable format for yolo object detection model(CNN) training• Worked closely with a team of research scientists and engineers to fine-tune object detection models, significantly improving their performance and accuracy for real-time applications in autonomous driving systems.• Ensured high data integrity by rigorously validating label accuracy against established benchmarks, leading to a 30% improvement in labeling accuracy. Show less

• Data Preparation for Deep Learning: Engaged in gathering, cleaning, and modifying data for the pretraining and fine-tuning of deep learning models• Research on Biological Language Models: Conducted research on various large-scale biological language deep learning models, including ESM2, Ankh, AlphaFold 2, ProTrans, OmegaFold, and PaddleHelix, addressing client demands and evaluated model performance using accuracy rates• Study of Diffusion Models for Geospatial Data: Investigated… Show more • Data Preparation for Deep Learning: Engaged in gathering, cleaning, and modifying data for the pretraining and fine-tuning of deep learning models• Research on Biological Language Models: Conducted research on various large-scale biological language deep learning models, including ESM2, Ankh, AlphaFold 2, ProTrans, OmegaFold, and PaddleHelix, addressing client demands and evaluated model performance using accuracy rates• Study of Diffusion Models for Geospatial Data: Investigated various diffusion models for segmenting geospatial data, such as RSPromter, SAMRS, and Geospatial, enhancing data analysis and insights Show less

• Utilized Wind Terminal to extract financial reports and industry data for the new energy vehicle and semiconductor industries• Refined and prepared datasets using R and Excel, ensuring data accuracy for in depth analysis• Analyzed share performance and financial health of leading companies in both industries, synthesizing crucial financial and managerial insights• Consolidated and presented research findings, effectively communicating insights and knowledge gained from various… Show more • Utilized Wind Terminal to extract financial reports and industry data for the new energy vehicle and semiconductor industries• Refined and prepared datasets using R and Excel, ensuring data accuracy for in depth analysis• Analyzed share performance and financial health of leading companies in both industries, synthesizing crucial financial and managerial insights• Consolidated and presented research findings, effectively communicating insights and knowledge gained from various industry roadshows Show less