Mohammad Aprialdi Rizky Pratama

oversee DS projects in paylater platform, including: - behavioral score: revamped existing b-score using enriched dataset that helped to save NPL by 10% - transaction score: managed the development of transaction scores that helped to save NPL by 5% by preventing fraudulent disbursement - in house device score: managed the development of device score using SDK data points that could replace dependency to external vendor and extend the usage to other initiatives - whitelist score: managed the development of whitelist score using platform data to target good non-PL users on traveloka platform and improve approval rate by 8%

- developed credit scoring models using alternative data to help FIs reduce their NPL rate - developed income prediction models from SLIK reports data - represented DS function during the OJK license registration

- Initiated propensity scoring model for SG Paylater products which helped reducing marketing budget to 10% by targeting better customer - Spearheaded credit scoring model development which 0.4 GINI and helped the Risk team whitelisting 35% of total MTU for upcoming lending product - Built a credit score for small ticket-size loan using user’s behavior data on OVO platform with AUC 75% and KS 0.35 - Built custom collection score to help prioritize debt collection effort which helps increase repayment by 13% - Built a sampling model that can capture desk collector’s poor quality performance with sensitivity up to 88% - Developed an algorithm to optimize field collector’s route by using machine learning models which resulted in increment of repayment amount by 18% and reduce man hour cost by 10% - Collaborated with local and regional team leaders and acting as a local representative in various discussions - Managed and oversaw 3 teams working on 6 verticals

Working on several machine learning projects, like product recommendation, product clustering, fraud detection system

- Conduct research of keyword extraction algorithm and build a keyword extractor for the news article in the website using Scala - Build a machine learning model to predict user demography based on his reading behavior using R - Build a recommender system based on reader’s interest and behavior using Scala and Spark - Devising mathematical equation for modeling user interest towards specific topic - Build a Named Entity Recognizer for Bahasa Indonesia model using Python and Spacy