Accelerating Machine Learning Engineering at Capital One
Capital One’s next generation of machine learning engineers
Senthil Kumar, Principal Scientist in Machine Learning at Capital One, has worked on various aspects of machine learning throughout his entire 17-year career. In his day job, Senthil focuses on everything from research — often convening workshops on ML in financial services at conferences like AAAI, KDD, and NeurIPS — to applying cutting-edge machine learning techniques to real-world business problems like fraud detection.
But one of the aspects Senthil most loves about his job is teaching associates across Capital One about machine learning. That’s why he was so excited to help build the curriculum for Capital One’s Machine Learning Engineering Training Program (MLETP) — a new program as part of Capital One’s Machine Learning Engineer job family — to assist software and data engineers shift to a career in machine learning engineering. “I'm happiest when I'm teaching. Creating the MLETP curriculum was so enjoyable — I was in a total state of bliss,” noted Senthil.
Helping Engineers Hit the Ground Running with ML at Capital One
Forty learners from all across Capital One joined the first cohort of the MLETP, an in-depth, 160-hour program that includes both instructor-led and self-paced content. Associates in the program receive hands-on experience using both AWS and Capital One machine learning tools, systems, and processes.
The curriculum primarily comprises four major focus areas: ML algorithms, engineering/MLOps, industry tools and Capital One tools and platforms. It covers everything from supervised learning, unsupervised learning, Spark, Dask and feature engineering to bias detection, performance metrics, parameter tuning, model validation, deploying scalable end-to-end pipelines, A/B testing, orchestrating MLworkflows, and performance monitoring.
Associates who participate in the program walk away with practical, hands-on skills and experience in machine learning engineering, including:
- A Coursera certificate and a portfolio of 4 hands-on projects
- Practice applying new knowledge to problem sets modeled on real projects
- Hands-on experience using Capital One data, systems, and frameworks
- Networking opportunities, mentorship, and leadership discussions
- The opportunity to transition to a career path in machine learning based on open job roles and discussions with their manager about their career path and goals
The program is also part of Capital One’s broader effort to formalize the machine learning engineering job family, where ML engineers can grow their skills and their careers while working on meaningful, challenging problems.
Learning to Combine Business Acumen with Engineering Techniques
Teresa Hardin, a 30-year software engineering veteran joined the program to become a better developer – and keep pace with advances in emerging tech — by marrying business and domain knowledge with technical know-how in coding and software engineering.
"I love software engineering. I can lose myself in it. I could work on it and lose track of time,” said Teresa. “Once I started learning about machine learning, the same thing happened. It's like solving a puzzle — and there's a business aspect to it, which I have a history in. So, you use a lot of your domain knowledge as well as your technical knowledge, and it's exciting what you can do with it. I found that passion again in machine learning.”
Some of the most valuable skills Teresa picked up in the program were learning how to get more sophisticated with data analysis, developing and tuning models, as well as implementing an ML model, how to monitor it, and how to make it work efficiently.
Building New Skills to Stay on the Cutting Edge
“I was inspired to gain machine learning expertise after hearing leaders in the company discuss the strategic importance of machine learning to Capital One several years ago — about how it was going to help us achieve things we never thought imaginable for both our customers and the broader business,” said Pradeep Raghunathan, a software engineer in Capital One’s Risk Technology group and a member of the first MLETP cohort.
Pradeep first joined Capital One as a contractor in 2016, and became a full-time associate in 2017. With a background in programming and algorithmic development, lifelong education and skills development is both a personal passion and a professional goal of his. After joining Capital One, he enrolled in a Masters Program in Information Technology at Virginia Tech, where he started to learn some of the building blocks of data science and machine learning. He next sought a deeper understanding of machine learning algorithms and how Capital One applies them to specific use cases.
The MLETP was a way for Pradeep to bridge the gap between surface-level research about machine learning and the very dense academic textbooks typical of advanced degree programs.
The program gave him more tangible knowledge of deploying machine learning from inception to production, all on Capital One’s own homegrown, enterprise-wide tools, services, and platforms.
A Collaborative Community of Instructors and Learners
For Ruikang Lan, a data analyst within experience in Capital One’s Card business and Tech College, the program was a way to expand the scope of her knowledge of machine learning — giving her a much broader perspective in areas she hadn’t been exposed to before, like platforms, packages, workflow, orchestration and other new topic areas.
What really made the program sing for Ruikang was the people at the heart of it — the support and collaboration of instructors, colleagues, and her manager, who encouraged her to participate to advance her skillsets. She added, “I think the MLETP goes past the limitations of other virtual boot camp/accelerated learning programs, because it’s really all about building a community of like-minded people working in the same company, in the same unique culture, who all want to grow their skills in this field.”
Maryam Esmaeilkhanian, a principal data engineer on the Risk Tech Consumer Credit Risk Management team and another MLETP cohort member, agrees with Ruikang. Apart from the technical skills she acquired, Maryam also learned about teamwork, collaboration, and mentorship through the program. “It was a great way to learn about other interesting machine learning work across lines of business, and to make new personal connections. I got to know more about the different kinds of technical problems experts are working on across the company, including lots of career insights.”
Today, Capital One is expanding the MLETP program to additional groups of learners, and is preparing to kick off its third cohort in the fall of 2022. Discover more about each of the associates featured in this piece in stories on the Capital One Careers Blog.
Learn more about tech and machine learning job opportunities at Capital One.