Software engineer with hands-on experience on
1. Deep learning model training and finetune
- Train a customized CNN model for nearly real-time broadcast football scoreboard recognition.
- Finetune human action recognition model for wedding video action recognition.
- Finetune content-based article-music retrieval model for article background music recommendation.
2. AI solution development and deployment
- Build AI microservice to do inference with custom models implemented by PyTorch, Scikit-learn and OpenCV.
- Provide AI solution assessment and PoC via studying papers and customizing models.
3. Data engineering and analysis
- Build data pipeline to collect advertising tracking data.
- Provide data analysis summary and visualization to the team to optimize advertising metrics.
- Automate anomaly detection to detect abnormal advertising metrics.
1. Go Agile : As a team member has experienced Scrum and Kanban.
2. Develop data pipe to compile new navigation map feature
- Develop global features such as map label placement, truck routing restrictions.
- Develop navigation input methods such as full-text search, pinyin search and next valid character search for Thailand, China and most of EU and NA countries.
3. Enhance map data analysis and anomaly detection
- Build web-based service to provide map data statistics and visualization to help the team get insights.
- Develop test cases to highlight map data anomalies.
1. Design and develop logistics web platform to support business customers : Wilmar International Ltd. and AMIGO Logistics Corp to make better logistics plans and decisions.
2. Survey and implement algorithms like metaheuristics to optimize factory assignment and routing plan.
3. Analysis and develop crowdfunding web platform for 104 Corp.
Chinese, English (moderate level)
Kaggle: Google Landmark Recognition Challenge, 195th of 483 https://www.kaggle.com/c/landmark-recognition-challenge/leaderboard
Read popular science book, watch movie and play game