Joseph Phang_OCBC Bank
A day in the life of…

Joseph Phang, Junior Risk Data Scientist

Joseph is a Junior Risk Data Scientist with OCBC Bank. He obtained his Bachelor of Accountancy and Business Management from Singapore Management University (SMU) in 2018.

Data science is a multi-disciplinary field that uses processes and algorithms to extract actionable insights from data. Graduating without a background in computer science did not deter me from obtaining this exciting in-demand role.

Specialising in credit risk, there is no typical day for me as a Risk Data Scientist. Activities are heavily dependent on project phases, ranging from development to project management. The week can be spent from debugging scripts to communicating concepts to project partners. For context, I am currently assisting in the System Integration Testing (SIT) phase of the bank’s first credit risk scoring model powered by machine learning.

8:30AM:

Mornings are usually the most impactful part of the day. Correspondence from IT project partners the night before may contain reports on connectivity errors between systems or new system requirements or infrastructural limitations. With these, I will craft an impact assessment after communications with our partners, and subsequently update my manager. Brainstorming ensues to decide what changes need to be put in place. For example, we may need to build a new module to receive eXtensible Markup Language (XML) messages from other systems, given the limitations on the initially requested JavaScript Object Notation (JSON) messages.

11:50AM:

Time for a breather with colleagues from different specialties. Lunch is a great opportunity to learn about developments in the bank outside my area of expertise. As a data scientist, I need to understand how my colleagues’ work can impact intricate interactions with our predictive model, so it is always great to hear from people across the bank.

1:00PM:

This is the time to check in with our project partners, which may include business analysts, IT infrastructure specialists who build ‘bridges’ between systems, or administrators who oversee systems performances. I also get to have fun playing detective, troubleshooting the script to apprehend errors. Should there be no hiccups in project delivery, stress-tests are done to ensure the model’s robustness. This includes creating a supplementary programme to direct a large number or variations of instances into our model to test the model’s resilience to high and varying traffic.

Depending on the new requirements that might come up throughout the day, there can be a variety of other tasks to prioritise and undertake. This is where creativity melds with technicalities which is to experiment and try out different strategies.

4:00PM:

If data science models are like paintings, knowledge is analogous to the different brush techniques. The more one knows about the underlying mechanics, the more ways one can create masterpieces. Once the experiment is concluded, I will consult with my manager on the underlying math, workings behind algorithms, or system structures to understand the model better. I appreciatethe open environment I work in where I am encouraged to ask questions. My manager is constantly willing to impart his knowledge, which motivates me to grow and learn without any boundaries.A year on from being new to data science, I have picked up programming languages such as Python and SAS, knowledge on statistical algorithms and project management skills.

6:00PM:

Towards the end of the day, we are up for some laser tag action, hockey or futsal. Before leaving, any development work is wrapped up and a major model experiment is triggered to run overnight. Time for the machines to work while we take a break!