The most in-demand AI skills at JPMorgan Chase, Citigroup, Capital One

Before artificial intelligence can eliminate jobs on Wall Street, it must first create them.

Posts for AI-related positions in banks including JPMorgan Chase, Citi Group And Capital one jumped 49% this year from 2025 to 139,819 registrations, according to an analysis from corporate hiring data firm Draup that was provided exclusively to CNBC.

The fastest growing area is a group of skills involving AI agents, according to Draup, which scrapes data from public job postings and platforms including LinkedIn. For example, references to agent orchestration, or the ability to design agents working together on a task, jumped 1,721% this year.

“It’s arguably the hottest skill on Wall Street,” Draup CEO Vijay Swaminathan said in an interview. “It’s a huge opportunity; they need people who understand data and who understand AI and where to put it.”

The job postings show that Wall Street banks are moving beyond chatbots and moving to the next phase of their AI strategy, one that has implications for executives, employees and shareholders. To deliver on AI’s promise to increase productivity and automate repetitive tasks, banks are looking to a future filled with armies of agents handling an increasing share of work.

While a previous AI hiring wave was dominated by engineers and data scientists who built models or adapted them to company data, the boom has expanded to include people tasked with integrating AI directly into lines of business.

Deploying AI within a financial institution often requires bringing together several specialized agents: one to inspect raw data, another to analyze a document and a third to verify regulatory compliance, for example.

The workers involved in this process, often called forward-deployed engineers, need a combination of technical capabilities and domain knowledge of a specific company or function, from trading desks to back-office operations and human resources, according to Swaminathan.

“A business is very complex,” Swaminathan said. “Sometimes these complexities are visible, but often they are hidden. It takes a lot of time even to automate a simple process.”

For example, creating a team of agents to automate the approval of a company’s vacation requests creates a web of edge cases and specific exemptions, he said.

The agent orchestration skill is particularly relevant for forward-deployed engineers because their job is to determine which agents are needed, what each does and what technology to use, the Draup CEO said. It also involves deciding when human supervisors should be involved, he said.

Other in-demand skills related to AI development involve understanding the tools and techniques that give agents the ability to get things done.

References to LangGraph, a framework for building multi-step workflows, jumped 679%, while LlamaIndex, which helps connect AI applications to data, increased 291%. References to retrieval augmented generation, or RAG, a technique for feeding AI models with information from company databases, soared 259%.

However, beyond technical abilities, emphasis is increasingly placed on so-called soft skills.

“Our analysis shows that there is a renewed interest in soft skills like problem solving, creativity, the ability to ask difficult questions, assertiveness (when it comes to) a deeper understanding of processes,” he said.

Other areas of growth within AI include those responsible for creating guardrails around emerging systems.

References in job postings to “responsible AI” jumped 657% this year, according to Draup, while those mentioning AI governance and risk management jumped 394% and 359%, respectively. Security teams also work to prevent third-party tools or external model connections from creating systemic vulnerabilities.

Governance-related skills now account for more than 16,000 references in Draup data, nearly double the approximately 8,400 related to training, deploying and running models.

“We take a lot of care to ensure that the third parties we use in these products don’t go rogue from a cybersecurity perspective,” he said.

Generative AI and agent roles are generally better paid than technology roles elsewhere in finance, with generative AI managers receiving a median base salary of around $190,000, according to Draup.

Despite higher salaries, filling these specialized positions remains a challenge, Swaminathan said.

To bridge this gap, big banks are relying heavily on internal reskilling programs to train existing developers and domain experts, he said. The development itself will create ripples of ripple effects: JPMorgan CEO Jamie Dimon has spoken of “huge redeployment plans” as AI takes over more work.

“I think the more we prioritize soft skills with the right amount of hard skills, people will adapt and learn,” Swaminathan said. “It’s a very exciting time for good talent.”