Thanks for giving the opportunity.
I have 13 years working experience.
The first two years I was a software engineer to implement an ERP system on Windows platform using .Net and SQL server.
Then I moved to [xxx] and work in [xxx] till now.
My first two years in [xxx] is to implement a customer segmentation system to support the retail bank marketing activities. The system extracts large dataset containing personal and financial info from mainframe, then classify the customer to different segmentation basing on pre-define criteria.
Later I joined [xxx] investment bank – Finance IT on 2007 and worked there for 8 years.
In the 8 years I was involved and leading many regulation and compliance projects from initiation to rollout and even production support. While the most complex and challenge project is called CORB3 (common reporting for B3). The project targets to reduce 140 bn $ RWA by using Basel3 recommend counterpart risk measurement approach such e.g. IMM model and SA-CCR approach. The implemented system collected and transformed data from front office / back office systems, and external data source, providing adjustment platform, Counterparty Risk Weighted Asset calculation and Risk calculation Engine(FERMAT) customization.
Last Sep, I joined [xxx] Group Big Data Service department as development specialist and also taking the scrum master role.
As a development specialist, I am involved in the design and implementation of the Data Lake platform, two data science applications: a customer classification system and a financial product recommendation system
The Data Lake platform includes data collection tier, data management tier and data consumption tier by vertically, by horizontally, it also includes info lifecycle management layer, metadata layer, security and governance tier. The Data Lake is implemented with Lambda architect via batch and real time Bigdata technologies.
The customer classification system targets to make tags on a customer on different dimensions such as financial and risk acceptance rating. It currently adopts supervised learning model – random forest taking 50000 training data and 20000 testing data to train the model, and used the trained model to predict and classify the all customers and new customers.
The financial product recommendation system takes the customer classification system result and the financial product features as an input , using random forest the train the model and predict and recommend products to potential customers.
The two Data science application is implemented via Spark and SparkML, Cassanndra. And we have plan to adopt ensemble methods with different supervised and unsupervised machine learning models to improve the predict accuracy.
Besides those product implement stuff, as a scrum master, I am leading the agile practice and Devops tooling adoption for the whole GBDS department. Improving the agility maturity from traditional scrum to scrum-water-fall and then scurmban, from single scrum team to scale agile with scrum of scrum approach, improving the productivity by adopting Devops pipeline with JIRA, Bamboo, Git, Nexus, Sonar and docker.