AI Jobs in Finance: The Roles, What Each Owns and Who Hires for Them

AI jobs in finance come under a handful of titles: the machine learning or AI engineer who builds the systems a fund's researchers use, the quant researcher who uses them to find a trading edge, the data scientist who builds and tests the tools an asset manager invests with, and the forward deployed engineer who builds AI inside someone else's operation.

Hedge funds and trading firms, asset managers, the big banks, the AI labs that sell to them and a few buyout firms all hire for these seats.

The title on a posting matters less than one question: what will you own? A trading idea, the system it runs on, a tool other teams rely on, or a build you hand over once it works. The answer decides the degree a firm looks for, the interview it runs and the pay it prints, and in most published hiring bars, code and statistics come before finance.

What AI jobs are there in finance?

These are the AI seats that finance firms describe in their own postings and careers pages, by title, with examples of who hires for each.

Sorting the titles by what they own

Read the rows by what each seat is responsible for, and four groups appear:

• The idea. Quant researchers and machine learning researchers own the hypothesis and the model behind a strategy.
• The system. Machine learning, performance and applied AI engineers own the code, the training runs and the tools the ideas run on, and so does a bank's AI platform team.
• The tool. Data scientists at asset managers and buyout firms, and AI engineers in compliance and operations, own something another team relies on to invest, check or serve clients.
• The deployment. Forward deployed engineers own a build inside someone else's operation, and the job ends when that team can run it alone.

The quant researcher and the machine learning engineer fall into the first two groups and sit at the same firms, which is why the next section takes them apart.

Who hires, and at what scale

The seats also come in very different numbers. A bank the size of JPMorgan has counted its AI specialists in the thousands, while at a fund or a buyout firm the AI team can be a few dozen people working directly with the investment or deal teams.

2,000+

AI and machine learning experts and data scientists at JPMorgan

2023 shareholder letter

About 60

Engineers and data scientists on one BlackRock portfolio-tech team

December 2025 profile

50-plus

Data scientists working with Blackstone deal teams

2023, per its head of data science

20

Size of the Balyasny Applied AI team at its start

late 2022, per an OpenAI case study

Quant researcher vs machine learning engineer

The two seats easiest to confuse sit at the same funds, write code in the same languages and lean on the same statistics. The firms' own pages still put a clear line between them, and it runs through ownership.

The idea and the system

Two Sigma's careers page describes quantitative research in two sentences: "Develop inspired ideas through scientific thinking, curiosity, and precision. Systematically test and expand hypotheses to shape company strategy." The same page credits the other side in one line: "Our tools and platforms are custom built by top engineers, enabling the creativity our researchers are known for."

D. E. Shaw's 2026 posting for a quantitative analyst describes the day in the same terms, running from "conceiving new trading ideas, formulating them into systematic strategies, and critically evaluating their performance." Its quants "apply mathematical techniques and write software," so the researcher codes too. The difference is what the code is for.

The engineer's work is the system underneath. A 2026 Citadel Securities posting built its machine learning engineer's job around a deep-learning library that "will empower 100+ researchers to iterate faster." The researcher's output is a strategy; the engineer's output is what strategies are built and run on.

How the two tracks are hired

The firms hire for the two seats on separate tracks, and their own pages show how far apart the tracks run.

Research track

• Citadel: a first round on programming, research ability and algorithms
• Two Sigma: open-ended data analysis, coding, and statistics or the research field
• D. E. Shaw: the top students from maths, statistics, physics and computer science

Engineering track

• Citadel: a first round on coding, data structures and algorithms
• Two Sigma: algorithms and object-oriented design, sometimes systems design
• D. E. Shaw: a degree in any field, plus an extensive software background

Pay is printed separately too. At D. E. Shaw, three postings from the same year cover the quant, the machine learning researcher and the applied AI engineer, each with its own bar.

Where the line blurs

The line is clear at the ends and blurred in the middle. Between the two seats sits a third, the machine learning researcher. At Jane Street, machine learning researchers invent and reinvent "the methods and models driving our trading strategies," and at D. E. Shaw they work with quantitative researchers and developers "to take successful ideas from initial exploration toward real-world application."

A quant title can also carry an engineer's toolkit. Man Group's 2026 posting for a quantitative researcher in AI and machine learning, in its Shanghai research team, asked for hands-on experience fine-tuning large foundation models, for a job built on using language models and alternative data to find new trading signals.

Even when a machine proposes the idea, the review splits along the same line. At Man Group, an investment committee judges an AI-made signal's economic rationale the way it would judge a human's, while technology teams review the code and test it.

AI engineers at hedge funds and trading firms

The engineering side of that line has its own profile. It follows the seat from a first project on company filings through the firms that hire for it, the toolkit the 2026 postings name and the code-first interview loops, and it covers what changes as agents start taking a researcher's first pass. Read it if you write software and want to know what building for researchers involves: what AI engineers build at hedge funds and how the funds hire them.

Forward deployed engineers: vendor, bank and buyout seats

The forward deployed engineer is the title on the map whose meaning shifts most with the employer. The profile traces where the title came from, sets out what a twelve-week engagement with a bank's financial-crimes team could look like, and compares the skills, pay and travel of the vendor, bank and buy-side seats. Read it if you want production engineering with a customer in the room: what a forward deployed engineer in finance does and who hires.

The asset manager's data scientist

At an asset manager the AI seat builds tools rather than trades, and the profile takes the world's largest asset manager as its case. It maps the teams there that hire for it, from the central lab to compliance, works through grading a compliance assistant question by question, and sets the firm beside three other large managers. Read it for the most measurement-heavy version of AI work in finance: what a data scientist builds at an asset manager.

What do AI jobs in finance ask for?

Across the map, the requirements come back to three things, and only one of them is finance.

Code that runs in production

Python is the constant. The second language follows the stack: CUDA and C++ where the funds need speed on GPUs, Java and Kafka in a bank's systems, SQL wherever the data sits. Morgan Stanley's 2026 posting for an engineer in its research technology team, for one, wanted "Strong Core Java, SpringBoot, Spring AI" next to Python.

What the postings share is production: code that other people depend on.

Proof that the model works

The requirement that recurs across seats is evaluation: building the test before anyone relies on the model. D. E. Shaw's applied AI engineer posting opened its duties with "Build, evaluate, and deploy." BlackRock's AI lab asked its data scientists for "benchmark construction, metric design, and statistically sound measurement of quality, safety, and reliability." Anthropic wanted "evaluation frameworks" from its customer-embedded engineers, and JPMorgan's private-bank AI engineer works inside "guardrails, evaluation, observability."

For a candidate it is the most portable skill on the map. A project that shows its test set, its score and where it failed reads the same to a fund, a bank and an asset manager.

Finance as a habit more than a credential

Finance knowledge is a plus in most engineering seats and a requirement in a few. BlackRock says a large majority of its engineers have no formal background in financial services, and Man Group called investment experience a plus but not required for its AI researcher. Morgan Stanley, by contrast, listed "Finance business understanding" among the requirements for its research technology engineer.

What carries over from finance is a habit. Vinoo Ganesh, whose company sells to hedge funds, banks and private equity firms, put the rule for all of them in one line: "numbers have to be right, and someone has to be able to show why they are right."

Which AI jobs in finance take graduates?

Most seats on the map ask for years of shipped work first. A few have a door from campus, and a couple have one from the business side.

Campus doors

Three kinds of employer on the map have one:

• Funds and trading firms open through internships. Jane Street's machine learning engineering internship puts students on real projects and does not expect a finance background, and Millennium listed a 2027 applied AI internship.
• Asset managers open through programmes. BlackRock takes technologists into a nine-week summer internship or a two-year analyst programme, sets a coding test for its software engineering and analytics applicants, and runs a separate internship for quantitative master's students.
• Deployment opens at Palantir, which hires graduates straight into its customer-embedded engineering role and promises that a new graduate "will not be handed a ticket queue."

A door from the business side

Two seats on the map were written for people who come from the business rather than from engineering:

• Citi's AI business analyst, the map row that swaps engineering years for business analysis or consulting.
• EQT's Motherbrain Labs, the buyout data science row, whose 2023 posting called the job "about communication and collaboration, not arcane computer science wizardry."

A structured self-check of what interviewers expect a candidate to know about AI lives on the interview readiness tool.

Are AI jobs in finance new?

Some are a decade old or more. The buy side built data and AI teams well before 2026 brought a wave of new titles, and the difference tells a candidate where each kind of opening sits.

The older rows are research and data science teams inside investment firms, built to help investors decide. The 2026 rows are about deployment: putting engineers inside a bank department, a portfolio company or a client until AI runs in production there, and that is where the newest titles sit. For a candidate, the older teams are small and close to the investors; the newer seats want engineers who can ship inside someone else's systems.

The bottom line

AI jobs in finance keep adding titles, but they still sort by the question the opening asked: what will you own? Pick the thing you want to answer for, whether an idea, a system, a tool or a build, and the title follows from it.

Whichever you choose, the published hiring bars ask first for code that runs and for proof that a model works. Finance knowledge helps, and the habit of making every number defensible helps more. The wider library of guides covers what the banks and buyout firms have built for themselves.