Gojek

• Developed and deployed an image similarity model to automatically flag fraudulent claims, resulting in the cost savings of $3k per week. The model achieved an AUC of 88.7%, and a 30% recall at an FPR of 0.25%.
• Maintained a daily human-in-the-loop fraud detection pipeline that detects GoFood merchant account takeover using keyword detection with a false positive rate of 1%. 
• Developed an NLP classification model to classify 11 million bilingual free-text retail items into 12 categories with an F1 score of 94%. Sales team used these classifications to create targeted messaging for product upselling.
• Built and monitored the performance of a production time-series model that predicts the weekly activity status of ~10,000 merchants with an F1-score of 90%. Marketing team used the result to reduce merchant churn.
• Built a random forest model that differentiated photos of premium restaurants and run-down eateries with an accuracy of 84% for classifying leads into high-, medium- or low-quality.

Data Scientist

• Performed a weekly merchant segmentation based on ~10,000 merchants on app activity using their variety, intensity, and frequency of actions on the app. Marketing team used this to personalize product communications.
• Planned, executed, and presented findings A/B testing on the effectiveness of new features such as self-onboarding materials and merchant loyalty program before rollout.
• Built and maintained dashboards for north-star metrics like monthly active users and Net Revenue Retention for management, product and business teams.

Data Analyst

• Instructed a 8-lesson data science bootcamp on hands-on machine learning algorithms for Heicoders Academy.
• Completed 50% development “Hive for Big Data” course on Educative and 
• Created 50+ pieces technical data science content of ~1000 to ~3000 words on various topics, including Generative AI, data science, Python and SQL.

Freelance Data Science Creator

Kuprion Inc

• Collaborated with a multi-disciplinary team to plan, run and analyze tests on a nano copper to fulfill customer requirements. 
• Prototyped and revamped company website using Figma, SquareSpace and CSS.

R&D Assistant

Fave

• Automated daily reports generation using VBA, reducing the lead time by half. 
• Critically analysed the performance up to 8,000 Fave Deals and merchandised top-performing Deals on Fave web and app platforms.

Growth Planner

Nanyang Technological University

• Investigated the knowledge transfer of engineering undergraduates.

Research Assistant

Xfers

• Automated the categorisation of inbound leads with Google Script
• Engaged in search engine optimisation for Xfers' landing pages

Marketing Executive

💡 Follow me for data science content + updates! 

🧑‍🏭 I'm a chemical engineer turned data scientist. I self-learnt everything I needed to become a data analyst/scientist through online courses. 

🤖 At work, I use machine learning and deep learning techniques to fight inappropriate content on the Gojek Merchant App. I use techniques like computer vision, time-series analysis, and natural language processing at work.

I also write about data science and data analytics. I share tips on python, SQL, and Google Sheets that saves me time.

I enjoy sharing my learning journey and knowledge. Let's connect and learn together.

Personal site: https://www.travistang.com
GitHub: https://www.github.com/travistangvh
Twitter: https://twitter.com/_travistang
Medium: https://medium.com/@travis-tang
Enquiries: travistang.vh@gmail.com

Taplio

Travis Tang 's Linkedin Analytics

Travis Tang

What is Travis talking about?

Who is engaging with Travis