Seven Questions With...

Siqi Wang joined the SentiLink team two years ago to work with our synthetic fraud and identity theft fraud models. She brought a deep understanding of advanced statistical theory and programming to the data science team, as well as her passion for exploring the usage of machine learning technique incorporated with statistical knowledge such as nonparametric Bayesian models and topic modeling.

Both her teammates and SentiLink's partners benefit greatly from Siqi's dedication to improving our model performance, and her passion and open-mindedness in her work make an amazing difference. We're so lucky to have her as a SentiLinker!

What's your typical day like at SentiLink?

It’s hard to summarize a typical day of my work since we often have new challenges and many of them are unique. Solving these challenges is what I’m always excited about. However I can enumerate some common themes of daily routine:

• Finding ways to improve model performance: From new attribute evaluation to feature engineering, from model retraining to code structure optimization, we’re dedicated to improving the performance of our fraud models from multiple perspectives.?
• Reviewing fraud cases: We always believe that getting close to actual cases is necessary to build good fraud models. Reviewing different scenarios not only inspires me with new feature ideas, but also allows me to better understand the rationale behind each labeling decision.
• Directly supporting SentiLink's partners: As data scientists, we’ll answer partners’ questions about the scoring rationale or the model structure. To me, this is a helpful opportunity to understand what our customers care about. Also, it’s a chance to identify places for model improvement.

How did you get started in your field of expertise?

I started to get interested in mathematics when I was in high school. Solving mathematical problems and writing elegant proofs is very satisfying and gives a sense of accomplishment.. Hence I decided to major in mathematics when I was in college. In my undergraduate study, I found that I’m particularly interested in statistics & probability related courses, so I went to graduate school to focus on statistics.?

When I was a graduate student, data science was a newly emerging term. I heard it required interdisciplinary knowledge of stats, math, computer science and business. All these sounded challenging and interesting to me. At the same time, I dreamt of leveraging my domain knowledge to make a real-world impact one day. So data scientist seems like a great fit to my goal. Later I did a summer data science internship which helped me confirm that being a data scientist is what I wanted to pursue for my career.?

I became a full-time data scientist once I graduated.

What's the most interesting thing you've worked on at SentiLink?

I’ve done many interesting projects here, so it's hard to pick my favorite one. I’ll just pick a recent one that I like a lot.?

The project is to build a new product to detect fraudulent applications that use the information of foreign students and workers. This was a very challenging project since the behavior is easy to disguise. I started my research by looking for commonalities in this kind of fraud behavior. Then I worked together with our fraud intelligence team to polish the fraudster portrait in order to determine the detection logic. Finally I implemented the code for production.

I really appreciated this opportunity to be the main data scientist to develop this new product from end to end. At the same time, I enjoyed the collaboration with so many smart colleagues across the enterprise.

What's your Employee Superpower, a valuable skill you've carried throughout your career that contributes to your success at SentiLink in support of our partners?

Not sure if it’s a superpower but to me being open-minded is crucial throughout my career. It includes multiple aspects:

• Being open-minded to new technologies and knowledge. In the tech world, technology evolves quickly. It’s important as a data scientist to learn new things and think about how to apply new techniques to improve existing works.
• Being open-minded to feedback from colleagues and customers. This helps me to spot ‘blind spots’ in my daily work and encourage more efficient collaborations.

What advice would you give to someone applying for a role at SentiLink?

Demonstrate your top-notch skillset and show your passion for this industry. At SentiLink, we set a high bar for the quality of our work. We always work diligently to provide the best product to our partners and protect people from fraud. It would require not only good technical skills but also a deep understanding of the fraud domain. Fraud detection is sophisticated and we often encounter ambiguous scenarios which require perseverance and smart ideas to demystify the truth. We love to work with talented people who are passionate and willing to take ownership to solve these questions.

What's the very first thing you ever purchased on Amazon?

I just checked my account and found it’s an AmazonBasics drying rack. I bought it when I moved to the graduate dorm at the University of Chicago. At that time I didn’t super trust the product quality for something with a name containing ‘basics’ but it turned out great. It’s a durable and useful purchase I’d say!

What's your hidden talent, something that can't be found on your LinkedIn profile?

In the last three years I've gotten really into skiing. I travel with my own equipment to explore and enjoy different places. Within the last year I’ve been to Aspen and Jackson Hole. I hope to go to Utah soon.