October 15, 2026
Share on
Maria Stoica’s internship began with a research paper.
On one of her first days at Citadel, her manager handed her a paper on accelerating mathematicians with agentic AI. It became an important starting point for the project she spent the summer developing.
Maria joined Citadel’s London office as a Quantitative Research Intern on the Model Validation team. Her PhD research focuses on AI safety and validating AI behavior, and the internship let her apply some of that same thinking to financial models and market data.
Her project focused on the firm’s end-of-day market data, where she was tasked with identifying market frictions and inefficiencies within messy data to help ensure a more reliable end-of-day price.
For Maria, that meant taking a problem she understood conceptually from her academic work and learning how to solve it within a live financial system.
Maria’s problem involved separating genuine market behavior from data distortions.
Market data can contain anomalies for many reasons, and the team wanted to know which observations reflected genuine market behavior and which ones distorted the pricing process. Maria’s project examined ways to detect those frictions and improve the information flowing into the firm’s end-of-day systems.
Jie Ding, who leads the Model Control and Governance Group, describes the problem as one of market efficiency. “We do see frictions in the market for different reasons. Maria’s task is essentially to help us find those frictions and fix them,” he says.
The scale of the data makes that difficult. According to Jie, the team has access to a large volume of information, but the relevant inefficiency may exist in only a small portion of it. The research problem is therefore not just processing data but finding the meaningful data points hidden within vast troves of information.
Maria was delighted to discover how closely her academic interests align with the firm’s work. “The most exciting part of my day here at Citadel is applying my PhD research to real-world problems,” she says.
Maria also brought a perspective the team could use beyond her core research project.
Her background in computer science and AI led her to experiment with AI coding tools and APIs that could help the team work through information arriving from different parts of the firm. In some cases, those tools could help classify information or automate tasks that previously required more manual work.
She also consolidated information about tools, APIs and data-visualization platforms to make a handy reference for her teammates. “I’ve been very proud of being able to bring in my AI expertise to the team and help with automating day-to-day tasks,” she says.
Maria also introduced methods from her own field to her colleagues and tested where they might be useful to the team.
Jie says that was intentional, as interns are expected to bring a fresh perspective to the work. “Our philosophy of the internship is to give the interns a project that others would dismiss as impossible. We want to give them as much help and support as possible for them to disrupt what we already do and make it better,” he adds.
Within Maria’s first week, she was already pushing code and helping restructure how the team maintained parts of its codebase. She was struck by how much her input was valued from the beginning.
Jie describes the team’s culture more simply: Interns are treated like members of the team.
Maria had access to the same systems, tools, information and colleagues as permanent team members. “Maria has 100% ownership over her project,” Jie says.
That’s not to say that Maria was working in isolation. Far from it, in fact.
Maria and Jie maintained a regular dialogue about her findings, how to improve her implementation and which results warranted further investigation. The feedback helped her go deeper in the analysis while also becoming faster at identifying promising areas.
The project also brought her into regular contact with quantitative researchers and engineers to develop methodology, write code or access the data she needed.
For Maria, one of the most important parts of that environment was learning how quickly she could close her own knowledge gaps.
She came into the internship with prior experience in finance as well as her PhD research, but there were still areas of financial mathematics, internal systems and market context she hadn’t encountered.
Her team helped her navigate those gaps by pointing her toward the right documentation, people and resources. They also worked through bugs in code, methodological questions and cases where pricing results didn’t behave as expected.
It reinforced a lesson Maria would give to future interns: Understand what you know but be equally direct about what you don’t. “The biggest piece of advice walking in [on] day one would be to own what you don’t know and to be confident in what you do know,” she says.
She found that asking questions early helped her build context faster and understand how to complete a task and why it mattered. “It’s important to understand the bigger picture. Not only how do you do it, but why should we care about this project and why are we doing this?” she says.
By the end of the internship, Maria’s work had connected several parts of her background that initially seemed separate: AI safety, model validation, computer science and finance.
Her academic experience helped shape the way she approached the project. At the same time, the project forced her to think about those skills in a new setting, where the value of an idea depended on what it revealed in real market data and whether it could improve an existing system.
Jie hoped that Maria would leave having contributed something useful to the team while also taking ideas back into her own research and professional development.
For Maria, that alignment between the work and her interests was a big part of the draw from the start. “The thing that made me really excited about joining Citadel and, specifically, this team for my internship was how aligned it was to my research interests. It’s challenging, but it’s been very rewarding,” she says.
Her summer started with a paper about AI agents and portfolios. From there, it became a practical research problem: work through messy market data, identify what matters, bring her own expertise to the answer and keep refining it with people who understood the problem from different angles.
Independence and feedback together shaped the experience. She owned the problem but asked for help when she needed it, and kept sight of why the work mattered.
