Introduction

The financial world runs on data, but who does it truly serve? For Michelle Zakson, a Compliance Officer, Fair, and Responsible Banking at BMO, and a Master’s student in Data Science at Northwestern University, the answer is not always fair or equitable. Her mission is to challenge that. By applying econometrics, machine learning, and ethical AI, Michelle uses data to root out discrimination in banking practices and push for a more inclusive financial future.

We sat down with Michelle to learn about her path from economics student to fairness-focused data professional, and how she’s using her skills to make the industry—and the world—a little bit better.

Career Path in Data Science & Industry Entry

Michelle's interest in data science was sparked during her undergraduate studies at Lake Forest College, when she took an econometrics course.

“It’s basically a data science course,” she says.

“You do a project from start to finish. You model, evaluate metrics like area under the curve (AUC), and most importantly, use your own discretion to judge what’s working.”

Though she didn’t know it at the time, that experience laid the groundwork for her eventual transition into data science.

Her curiosity deepened at work. While she was doing mostly qualitative analysis, a colleague’s fully quantitative approach intrigued her.

“I saw how impactful that kind of work could be, and I really liked it,” Michelle recalls.

“That’s when I started thinking seriously about a Master’s in Data Science.”

But she didn’t jump in blindly.

“I spent a lot of time gathering industry knowledge, figuring out the tools professionals were using and how they were doing the analysis,” she explains.

That research led her to Northwestern, where the curriculum offered a blend of technical depth and business translation. And this was exactly the balance she was looking for.

Her first step into the compliance world was “pretty random,” as she describes it. Just applying to anything that sounded interesting. However, with her economics background and extracurricular experience with the Federal Reserve Challenge team at LFC, her resume stood out. That exposure to regulation and policymaking impressed her future employers. Later, at Wintrust, she built habits that became critical: curiosity, initiative, and a willingness to explore different software and tools.

“My manager at Wintrust saw that and recruited me to join BMO,” she says.

The transition from school to industry wasn’t without challenges.

“At school, everything is calculated and guided. At work, you have trust, independence, but also real-world consequences,” she says.

“You have to figure out how to apply your knowledge, and how to troubleshoot. That part was hard.”

Her solution? Ask for help. “I’m young and naïve. I ask all the time. Luckily, I work with people who are really knowledgeable.”

Social Impact and Data Ethics

So why stay in compliance and consumer protection for the long haul?

“This is what I know. I’ve been working in consumer protection banking my whole career, and I stayed because I believe it’s a good use of my time,” she says.

“There are so many inequalities in the U.S., and being able to use data science to help alleviate that—even just a little—is powerful.”

She’s driven by the belief that algorithms shouldn’t reinforce the biases of the world we already live in.

“Models determine whether you get a home or even an interview. However, these models are built on biased data. The world is already unfair,” Michelle says.

“Our job is to recognize those biases and work around them to make the world fairer, to create a real meritocracy.”

At BMO, together with her team, she validates and monitors models for customer interactions like loans and mortgages, while other teams also monitor internal practices like hiring. It’s about making sure that if you’re the right candidate, you get the job, regardless of your race, gender, or background.

Emerging Trends & Technologies in Fair Finance

In her work, Michelle uses a range of tools to uncover discriminatory patterns: Python, multiple linear and logistic regressions, odds ratios, relative risk measures, and even ArcGIS for geographic mapping.

“We map where mortgages are approved or denied, and check whether certain neighborhoods are disproportionately affected,” she explains.

Her team compares BMO’s lending practices with peer banks to ensure fair treatment of protected classes.

But the real magic, Michelle insists, happens before modeling even begins:

“Exploratory Data Analysis (EDA) is more important than model building. Garbage in, garbage out. If your data isn’t cleaned or normalized properly, you can’t trust your model’s results.”

The landscape changes every year, with new formats and structures in datasets.

“You constantly have to pivot and adjust. It will never be 100% accurate, but it’s the best thing we have,” she says.

“Take models with a grain of salt.”

One trend Michelle is especially passionate about is the rise of alternative data. Traditional models use credit scores and loan histories, but what if someone doesn’t have a mortgage or even a credit card? That doesn’t mean they’re irresponsible. Alternative data, like utility payments or peer-to-peer lending activity, allows banks to assess creditworthiness in new, more inclusive ways.

“It’s about fixing inequality. If someone’s worthy of a loan, they should get it.”

She also stresses the importance of ongoing model monitoring, especially to prevent discriminatory drift after deployment.

“We look back and make sure our models aren’t disproportionately affecting protected classes negatively. It’s not a one-time check, but a continuous process.”

Projects & Hands-On Experience

One project Michelle highlights involves publicly available mortgage application data. Anyone can access it. It includes loan decisions along with ethnicity, age, gender, and geography. Her team used this data to uncover disparities in approval rates across communities, and then mapped those results to visualize the trends.

“Every project shows disparities. It reflects the real world.”

But what happens when they do find something?

“We go straight to the business units. We’re not there to attack; they’re our partners. We show them the data and ask how we can fix it.”

Sometimes that means hiring more loan officers in underserved areas, changing marketing tactics, or updating internal procedures.

“Discrimination is inefficient in economics. If someone has demand for a loan and is qualified, but can’t access it due to bias—that’s a market failure.”

Staying Sharp and Giving Back

Michelle stays updated by subscribing to data science and engineering newsletters, watching documentaries, reading books, and attending conferences in finance and tech.

“You don’t need grad school to stay informed. There are free courses from Harvard, Princeton, Stanford, and many other universities. Professors explain things in ways that might finally click.”

To stay ahead of emerging trends, she taps into a variety of resources, including industry publications, expert networks, and AI tools like ChatGPT, Microsoft Copilot, and Google Gemini.

“They’re great for quick updates on what’s trending and what’s coming next. They also point me to supporting articles that help deepen my understanding.”

Here are some of her suggestions, more specifically (including links):

1. Documentaries:

   1. In the Age of AI

   2. Coded Bias

2. Books:

   1. Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy by Cathy O'Neil 

   2. Invisible Women: Data Bias in a World Designed for Men by Caroline Criado Perez

   3. Competing on Analytics: The New Science of Winning by Thomas H. Davenport & Jeanne G. Harris

   4. The Color of Law: A Forgotten History of How Our Government Segregated America by Richard Rothstein

3. Newsletters:

   1. Data Science Weekly 

   2. O'Reilly

Advice for Aspiring Data Scientists

Michelle’s advice to students hoping to work at the intersection of data and social impact is clear:

1. Learn Python – “Everyone uses it. Every industry. Especially for social impact.”

2. Master Machine Learning – “Structured, unstructured, semi-structured. Know how to communicate it clearly to business stakeholders.”

3. Build Data Visualization Skills – “Don’t just show raw numbers. Use correlation matrices, geospatial maps. Help people see the insights.”

And most importantly?

“Understand the bias in the world. Watch documentaries, read books. Do whatever it takes. Know the social problem before trying to solve it. This is how we create social impact.”

Passion, she says, is what separates a good candidate from a great one.

“It’s not hard to find someone with the skills. It’s hard to find someone who cares. Someone who’s serious about making a change will always produce better work.”

Michelle Zakson proves that data science isn't just about models and code. It's about the people those models affect. And with more professionals like her, the financial industry has a real shot at becoming more just, equitable, and inclusive.