◦ Web-App Development: Worked in a team to develop a Computer Vision based social media web-app, from

scratch to MVP in 10 weeks, following an Agile development methodology.

◦ Microservices Architecture: Designed web-application in Flask, integrated with a PostgreSQL database and

REST APIs for computer vision functions, which I deployed on Google Cloud Run.

◦ Performance Optimization: Used Redis Queues to create async API function calls in Flask and reduced overall

response time from 30 to 5 seconds.

◦ CV Model Design: Trained and tested 2 object detection models (R-CNN & YOLO) for aircraft detection on

high-resolution satellite imagery, and developed benchmark metrics for their performance.

◦ Data Pre-processing: Rectified topographical tilt (orthorectification) and lens distortion of 10,000 satellite

images, and used AWS SageMaker to establish ground-truth training data.

◦ Decision Recommendation: Computed benchmarks on company asset-monitoring algorithms and prepared

recommendations for object detection models to improve performance metrics.

◦ Data Science Conferences: Represented ByteDance AI Lab in international Data Science conferences (e.g.

KDD2018 London). Organized company welcome events for over 100 academics invitees.

◦ Coding Bootcamp: Developed legal framework and organized company sponsored coding bootcamp of over

500 international participants.

◦ Academic Brand Image: Designed and produced representative brand materials for academic partners (e.g.

brochures and website).