• Developed production-ready microservices for HubSpot and Salesforce data extraction using Flask, PostgreSQL, and Redis

• Implemented batch processing, API retries, connection pooling, and structured logging for reliable data processing

• Designed RESTful APIs with Swagger documentation, extracting 10+ CRM properties from HubSpot v3 API endpoints with paginated response handling, collaborating closely with managers on requirements and code reviews

• Built testing framework with pytest (unit, integration, edge case, performance), covering 8 API endpoints

• Deployed containerized applications with Docker across dev, test, and production environments

Designed an eye-tracking data acquisition system incorporating infrared emitters, a high-resolution camera, and MEMS mirrors with dual-axis oscillation.

Developed an algorithm to calculate the vector between Purkinje image coordinates and segmented pupil centers, enabling precise pupil-centered tracking for corneal reflection analysis.

Performed data augmentation on 1,416 dental X-rays using OpenCV, applying filtering and geometric transformations to enhance training datasets.

Applied the U-Net semantic segmentation model to eliminate interference of irrelevant information on caries recognition.

Introduced and compared different attention modules and trained YOLOv5 on training datasets to identify caries lesions.

Adjusted hyperparameters and activation functions, making the mAP@0.5 value reach 0.90 and accuracy reach 0.95.