The Co-op Close-up: AutoML and Fintech at UMF

SFU’s professional master’s program in computer science trains computational specialists in big data and visual computing. All students complete a paid co-op work placement as part of their degree. In this feature, we examine the co-op experiences of some of our big data students.

Btara Truhandarien completed a Bachelor of Computer Science from the University of Waterloo. He worked as a software engineer at Japanese e-commerce giant Rakuten for two years before joining SFU’s professional master’s program in computer science.

Can you tell us about UMF? What is it like working there?

UMF, short for Union Mobile Financial Technology, was established in China in 2003 and has become a strong player in the Chinese financial market ever since. The company powers much of the Chinese market’s financial transactions for consumers, enterprises of various sizes, and financial institutions by providing fintech and payment solutions. In 2015, UMF started its overseas expansion and now has two locations, one of which is in Vancouver, BC. The office in Vancouver, where I work, is an R&D branch and develops new technologies for the company. Due to the nature of researching and building cutting-edge technology, the branch has a good amount of liberty in its approach and solutions, while still interacting with the main branch in China to stay aligned with the overall mission and vision. The branch work schedule is project-oriented, and, for this year, we are focusing on building an automated machine learning platform (AutoML) for the company.

Can you tell us a bit more about machine learning and AutoML?

There are multiple steps within the development of a machine learning algorithm, also known as the machine learning life cycle. Briefly, those steps are data gathering, data pre-processing or cleaning, feature engineering, feature selection, model training and tuning, model deployment, and model maintenance and monitoring. A complete AutoML system aims to achieve the automation of all of these steps, except data gathering. This enables people of broad skill levels to create powerful and effective machine learning solutions to various problem domains.

What are your responsibilities in the project you are working on?

The project I have been working on involves building a drag-and-drop machine learning web platform. This platform, similar to Microsoft Azure, allows users to build machine learning models using their own datasets. Most of UMF’s clients are financial institutions and governmental organizations working with financial data. So a simple example of how the platform can be used is to create a model that predicts whether a customer will pay back their loans, based on possibly thousands of features.

As the main developer of the translation API layer, I was responsible for handling the data flow between the user-facing data and the data structures required to execute various user commands. It is also this layer’s responsibility to store any required metadata information and decide what kind of data users receive, when they will receive it, and how they will receive it. In conjunction with the other parts of the system, the platform we have built enables users to provide their own datasets to the system, explore the datasets’ statistical information, build experiments and models, and execute the experiments with various parameters.

How has SFU’s master’s program prepared you for your co-op work?

There is no better experience for learning complex problems than putting your own two hands to the problem directly. I find that the big data program at SFU highly encourages this through the course projects. I am fortunate to have worked on projects that are technically challenging like my capsule networks project or the job advisor model project which covered a variety of techniques and data sources. The projects cultivated my research, design, technical, and critical analysis skills - each crucial for building the AutoML platform. Without the hands-on approach of the program, my understanding and skills could not have grown as much as they have during my co-op.

Where have you seen your biggest areas of growth during co-op?

I felt my largest growth has come from the responsibilities and trust that I have been given. While I do work within a team, and there is somebody who occasionally helps me implement features, for the most part, I work on the API layer alone. As the main developer, I have the responsibility of navigating through technical difficulties. I am often challenged with open-ended design decisions such as structuring the data flow of the system. This drives me to always critically assess the decisions I make and carefully plan for the potential impact in the future. In a sense, I am not just a developer but also a system designer. This design skill applies beyond the system-level design, extending to the design of the implementation. For example, when I implement the solution in code, I am always self-checking myself rigorously by going through several decision points examining code testability, maintainability, flexibility, usability and more.

What are your most valuable takeaways from this co-op experience?

I feel fortunate to be working on a project that is as challenging and unique as AutoML, and am grateful for the experience gained from packaging it as a platform to be used by users. It is both challenging and innovative, and it is something I can be proud of talking about when I look for future work opportunities.

The second thing I feel most fortunate about is the unique experience of working within a cross-cultural team. Part of my team members are based in UMF’s China branch. This makes it tricky to communicate ideas and points across due to the obvious reasons of language barrier and time zone difference, but I have managed so far with the help of my team members and also by communicating through technical designs.

Overall, my work at UMF has been a pleasant and unique experience. The culture of trust and autonomy is truly beneficial in helping me grow on the technical side and develop my soft skills. I truly believe that the culture of trust within the company has enabled me to maximally utilize my capabilities and I look forward to my next work term here at UMF.

What do employers say about our students?

“Btara is a very capable, responsible, and supportive team player. He has done an outstanding job with his substantial knowledge and practical working experience. We are so grateful to have him on the team. SFU has high quality talent and programs that can help the company grow. We are very happy to work with SFU.”

— Carrie Li, Vice President at Union Mobile Financial Technology (Canada)