Jane Street internships

A trading interview is not a software interview, and Strategy and Product is not conventional consumer-product management. Jane Street’s roles share a trading business, but the output you own determines both your application and your preparation. [4]–[13][15]–[20]

Jane Street provides liquidity and trades across ETFs, equities, bonds, options and other markets. Its current client material says the firm operates on more than 200 trading venues in more than 45 countries, prices and trades more than 10,000 ETFs, trades equities in more than 45 countries, priced more than 25,000 bonds and traded more than $900 billion of bonds with clients in 2025, and actively prices options on roughly 3,800 tickers. These figures describe the scale and range of the business, not internship headcount or a promise that an intern will touch every asset class. [1][2]

The business is useful to understand because trading performance depends on more than a person deciding whether a price is high or low. A trading or client problem can require a model, clean data, fast and reliable software, exchange connectivity, hardware, operational controls, and communication across several teams. Jane Street explicitly describes the boundaries among trading, research and technology as porous. “Porous” does not mean interchangeable. Each function has a different output and a different failure mode. [1]

Quantitative Trading

The output is a decision or tool that makes trading better: a cleaner estimate, a useful model, a better way to reason about a market, a trading study, or software that supports the desk. The internal users are traders and adjacent researchers or engineers. The strongest applicant evidence is therefore not a generic finance résumé. It is a record of quantitative reasoning under uncertainty. Good examples include probability-heavy research, competitive mathematics, data projects where the candidate had to revise a hypothesis, game-theoretic work, or code used to test a real decision.

Suggested preparation: practice probability, conditional reasoning, expected value, estimation and market-making exercises while speaking through assumptions. Jane Street’s own Probability & Markets guide covers probability, expected value and making markets, and the firm says this is representative preparation rather than required prior knowledge. [16]

Quantitative Research

Quantitative Research interns work on longer experimental loops. The current New York page describes identifying market signals, generating and analyzing datasets, designing experiments, engineering features, building models, studying time series and productionizing useful findings. Jane Street says its researchers use methods ranging from linear models to deep learning depending on the problem and have access to large compute and data infrastructure. Most candidates have data-science or machine-learning experience; research experience is a plus. Python comfort and strong mathematical reasoning are explicit expectations. [5]

A strong research candidate should be able to explain an entire experiment rather than only the model. What was the hypothesis? How was the dataset generated? What was the baseline? What could leak future information? How did performance change out of sample? What did a failed result teach you? Who would act differently because of the result? That is closer to the role than naming fashionable models without explaining validation.

Trading and research overlap, but the center of gravity differs. Trading is closer to real-time pricing, risk and decision making. Research spends more time on experiments, data and signal validity. Jane Street’s own interview-preparation material groups quantitative trading and research around probability, expected value and market reasoning, but the role pages show different day-to-day outputs. [5][16]

Software Engineering

The user of the work may be a trader, researcher, another engineer, or an internal operations team. Strong evidence is therefore technical ownership under constraints: a distributed system that stayed correct under failure, a compiler or language project, a database or storage system, performance work with measured bottlenecks, a networked service, developer tooling that changed team productivity, or a technically serious product where the candidate can isolate their contribution.

Jane Street’s software interview guidance says candidates use a real programming language they know, not pseudocode. It focuses on writing working code, basic data structures and APIs, testing, edge cases and collaborative reasoning. The firm specifically discourages learning OCaml solely for the interview. [15]

Machine Learning Engineering

The distinction from Quantitative Research matters. A researcher may spend more time asking whether a signal is statistically real and economically useful. An ML engineer may spend more time making training, inference, data movement, tooling and reproducibility work at scale. A candidate with a strong systems project should explain throughput, memory, distributed-training trade-offs, failure recovery, inference latency or debugging. A candidate whose strongest work is experimental research may fit Quantitative Research better.

Data Engineering

The downstream user is often a researcher or trader who needs a dataset they can trust. This makes provenance and semantics as important as code. Strong evidence includes schema design, lineage, deduplication, event-time reasoning, backfills, validation rules, anomaly detection, monitoring and careful treatment of late or revised data. A notebook that generates a chart can demonstrate analysis; it does not by itself demonstrate the reliability work this role requires.

Hardware Engineering

The Hardware Engineer Internship is a narrower route for candidates with FPGA or ASIC depth. The current New York page asks for experience with an HDL or hardware-construction language and vendor tooling, plus a completed hardware project. The work is tied to high-performance systems and networking rather than general corporate hardware support. [9]

The convincing portfolio is concrete: architecture, RTL, simulation, timing closure, verification, synthesis, resource trade-offs, board or device bring-up and debugging. Finance knowledge is secondary. A candidate who cannot discuss timing, constraints and verification is unlikely to compensate with trading interest alone.

Trading Desk Operations Engineering

This is not generic back-office administration, and it is not simply Software Engineering with a different title. A TDOE may be closer to live system behavior, incidents, controls, operational state and coordination across desks. The output is often a system, investigation or process that keeps trading reliable. Strong evidence includes incident response, automation, data queries, process redesign and projects where the candidate had to coordinate technical and nontechnical stakeholders without losing precision.

Jane Street publishes a separate TDOE interview process: a take-home exercise, an initial problem-solving Zoom interview, a more technical Zoom interview, and a final round that may be virtual or in person. The firm says the exercises are meant to test learning style, attention to detail and the ability to work through unfamiliar problems. That sequence should not be generalized to other internships. [17]

Strategy and Product

This is not conventional consumer-product management. A project can involve an internal trading tool, market structure, a business process, operational infrastructure or a cross-functional decision. The user might be a trader, engineer, researcher, operations team or business lead. Strong applicant evidence is a project where the candidate found the real problem, gathered evidence from several stakeholders, made trade-offs explicit and drove a decision or launch.

The interview process is also distinct. Jane Street says Strategy and Product typically uses an online assessment, one or two video interviews and a final in-person round. The interviews test learning and problem solving rather than prior business frameworks. [19]

Institutional Sales and Trading

This route differs from Quantitative Trading. A quantitative trader is primarily reasoning about pricing, signals, risk and execution inside the trading system. Institutional Sales and Trading adds client objectives, relationship judgment and communication. The strongest candidate is neither purely social nor purely technical. Evidence should show numerical comfort plus the ability to explain complex information, handle pressure and build trust.

Tools and Compilers Research

This should not be treated as another general undergraduate SWE internship. The relevant evidence is graduate research with technical depth, ideally connected to a system or open-source artifact that can be discussed beyond the paper abstract.

Prioritize Jane Street if the actual work product appeals to you and your evidence matches it. Quantitative Trading suits people who enjoy decisions under uncertainty and fast feedback. Quantitative Research suits people who like longer experiments, data and model validation. Software Engineering suits builders who care about correctness, performance and abstractions. ML Engineering suits candidates who want to make ML systems work at scale. Data Engineering suits people who are unusually careful about what data means and whether it can be trusted. Hardware suits candidates with real FPGA/ASIC work. TDOE suits technical operators who like live systems and cross-team problem solving. Strategy and Product suits structured generalists who can move between technical, operational and market problems. Institutional Sales and Trading suits analytical communicators who want client exposure.

The main trade-offs are selection risk, specialized preparation and a fast feedback environment. High displayed salary should not decide the role choice. A $300,000 salary field attached to work you do not want is a worse target than a lower displayed field attached to work in which your experience is credible and your day-to-day interests are strong.

Ask who will use your work, who will review it, and how often you will receive feedback. For trading and research, ask how an experiment reaches a trading decision. For engineering, ask who owns the system after your internship. For Strategy and Product or TDOE, ask what authority you will have to change a workflow.

Get the return-offer criteria and decision timetable from the recruiter, then confirm the cash-pay convention and any housing or relocation support in writing. These are questions to resolve, not claims about standard Jane Street benefits or offer rates.

1. Pick the function before doing interview prep. Write down the actual output you want to own: a trading decision, research signal, production system, ML pipeline, reliable dataset, hardware design, operational process, cross-functional product decision or client outcome.

2. Build one work sample that resembles that output. For research, show the full experimental loop. For software, show architecture and failure handling. For data, show provenance and monitoring. For hardware, show verification and timing. For Strategy and Product, show a decision memo backed by evidence.

3. Use Jane Street’s official interview guides before third-party puzzle lists. Practice speaking through uncertainty and revising after feedback.

4. Apply when you are ready to interview. Do not infer a universal deadline from Strategy and Product’s September 30 planning date.

6. Before accepting, get the actual cash-pay convention, bonus terms, housing or relocation support, dates, office expectations, work authorization treatment and return-offer criteria from the offer or recruiter.