June 24, 2026
Share on
Artificial intelligence is changing how software gets written.
Tools that once felt experimental are quickly becoming part of everyday engineering workflows. Engineers can now generate scaffolding for systems, explore unfamiliar codebases and automate parts of the development process in ways that would have seemed improbable even a few years ago.
The pace of iteration is increasing, and the cost of experimentation is plummeting. But the most important shift may not be the tools themselves but how engineers use them.
At Citadel, engineers spend much of their time building systems that sit close to the firm’s core decision-making processes. Data pipelines, research platforms, trading infrastructure and analytics tools all shape how information moves through the organization. As AI expands what engineers can do, it raises new questions about how those systems should be designed and how engineering work itself evolves.
This moment is defined more by leverage than automation.
Much of the recent conversation about AI focuses on how quickly it can produce code.
Although that observation is accurate, it can also be misleading.
The hardest part of engineering has never been generating code. Designing systems that behave reliably, scale under pressure and produce meaningful results remains the far more difficult challenge.
AI shifts where time is spent. When scaffolding and repetitive implementation become easier, engineers can focus more on architecture, debugging system behavior and thinking through how software interacts with real-world constraints.
Andrew Janian, Interim Chief Technology Officer at Citadel, frames the change simply:
“AI can generate lines of code faster than any of us. That means writing code is no longer the bottleneck. The real value is deciding what should be built and how the system should behave.”
In many ways, AI makes the craft of engineering more visible because decisions about design, performance and failure modes matter more.
The work shifts toward reasoning about systems rather than assembling them line by line.
One crucial aspect of engineering work at Citadel is how closely it sits to the problems the firm is trying to solve.
Engineers build infrastructure that researchers use to explore ideas, tools that help investors interpret information and systems that allow large volumes of data to move quickly and reliably. These systems influence how quickly insights can be tested and how confidently decisions can be made.
That proximity tends to shape the engineering process. In fact, technical work rarely happens in isolation. Engineers often find themselves discussing data interpretation, workflow design or performance constraints with colleagues in research and investment roles. The goal is usually not just to make a system run faster, but to make it more useful.
Richard Lee, who leads Quantitative Development for Citadel’s Equity Quantitative Research strategy, explains this dynamic:
“You can have the smartest researchers in the world, but if they cannot test their ideas quickly, progress slows down. Our job as engineers is to make that process faster and more reliable.”
AI adds a new layer to that dynamic. As tools make experimentation easier, the questions engineers ask about what to build and how systems should behave become even more central.
For engineers, AI serves as another layer of tooling.
It can help navigate large codebases, accelerate repetitive work and provide quick starting points for complex implementations. In some cases, it can compress hours of exploration into minutes.
The effect is similar to other moments in engineering history when new abstractions changed how work was done. Higher-level programming languages, distributed systems frameworks and modern development environments all had similar effects.
Each shift removed friction from the process while introducing new kinds of complexity elsewhere. AI appears to be following a similar trajectory.
Although it’s becoming easier to write code in certain contexts, understanding systems and their behavior remains just as difficult.
If AI makes it easier to generate software, it also increases the importance of evaluating what that software does, as engineering judgment becomes more visible.
Questions about reliability, scalability and accuracy are still fundamentally human problems. Engineers need to understand how systems interact with each other, how data moves through infrastructure and how unexpected conditions might affect behavior.
Janian notes that AI’s impact depends largely on the engineer using it:
“These tools are leverage. If you already have strong engineering habits, they make you faster and more effective. If you do not, they can amplify the wrong things just as quickly.”
In environments where systems operate continuously and interact with complex markets, those considerations are especially important.
AI can assist in building software, but engineers still need to understand the systems they operate. If anything, AI makes that need more apparent
Like many organizations working with complex infrastructure, Citadel is experimenting with AI while paying close attention to how it interacts with existing systems.
New tools are introduced gradually, often with clear guardrails around how they interact with internal codebases, data and infrastructure. Engineers are encouraged to experiment and share feedback, while centralized teams work through questions related to security, reliability and operational risk.
That approach can sometimes feel slower than the pace of the broader AI conversation, but it reflects a simple reality. Systems that support markets require stability as well as speed.
Over time, the most useful patterns become part of the engineering workflow.
While AI probably won’t replace the need for engineers, it is more likely to change what engineering work looks like.
The role may continue shifting toward architecture, system design and interpreting complex outputs. Engineers may spend less time implementing individual components and more time shaping the systems that house those components.
In environments where technology influences real-world processes, that shift can be particularly meaningful.
Systems that once took months to prototype may take weeks. Ideas can be tested more quickly. Infrastructure can evolve more rapidly as engineers explore different approaches.
The underlying challenge, building reliable systems that help people understand complex problems, remains familiar.
AI is accelerating many aspects of engineering work, but the underlying questions remain the same. How do you build systems that are reliable, useful and capable of supporting complex decisions?
At Citadel, those systems sit close to the firm’s core activities. Engineers work alongside researchers and investors to design platforms so ideas can be tested, data can be understood and decisions are made with enhanced clarity and speed.
Over time, that proximity shapes how engineers think, helping them develop the commercial judgment that comes from seeing how systems influence tangible outcomes.
The emergence of AI tools increases the leverage engineers have within that environment. When experimentation becomes faster, and systems become easier to build, recognizing which problems matter and designing systems that solve them well may be the most valuable skills of all.
For engineers working in this environment, the opportunity compounds over time.
The tools are changing quickly, but the motivation behind the work stays the same: understanding complex systems, solving difficult problems and building technology that works.
