Goldman Sachs Interview Guide 2026
The Goldman Sachs interview process for technology roles is one of the more predictable in finance-tech, which sounds like good news. It mostly is. But “predictable” doesn’t mean forgiving. Goldman replaced in-person first-round campus interviews with HireVue specifically to expand beyond the schools they had historically recruited from, according to HireVue’s own writeup on the redesign. That’s a meaningful change in surface area. More applicants get in the door. The bar for making it through doesn’t move.
At LastRound AI, we’ve watched candidates prep for Goldman Sachs interviews using our AI interview copilot across dozens of live mock sessions. The patterns are pretty consistent: people over-rotate on LeetCode hard problems and under-prepare for the “Why Goldman over Big Tech?” question, which trips up a surprising number of otherwise strong candidates. More on that below.
Why the Goldman Sachs interview feels different from FAANG
The short version: Goldman is a technology company that has to explain that to people. The longer version involves 30 years of SecDB, a proprietary risk platform that Goldman engineers built before “fintech” was a word, and which is reportedly still one of the most sophisticated real-time pricing systems running anywhere. The firm has open-sourced parts of that work through GS Quant on developer.gs.com, a Python toolkit for quantitative finance. Worth reading before your interviews, not because they’ll ask about it directly, but because it tells you what kind of engineering culture you’re walking into.
What you’ll hear from interviewers is a variant of: “we are a technology company that happens to be in finance.” That framing is genuine. What it means practically is that the interview values overlap with Google or Amazon on technical depth, but the domain context matters in ways it doesn’t at pure SaaS companies. You’re expected to know what a trade execution looks like end-to-end. You don’t need a CFA. But “I’m not really a finance person” reads as lack of preparation, not refreshing candor.
The work-life balance question comes up constantly in the candidate prep sessions we run. Honest answer: tech hours at Goldman are not banking hours, but they’re not Google hours either. Expect something closer to a demanding product company during market events, quarterly closes, and regulatory filing deadlines. Most teams don’t have on-call rotations in the classic SRE sense, but when a trading system misbehaves before market open, people are expected to respond.
The Goldman Sachs interview stages, one by one
The full process runs four to eight weeks depending on whether you’re a campus hire or lateral. Here’s what each stage actually involves.
Stage 1: Online assessment (HackerRank)
Three coding questions, 120 minutes. You choose between a pure programming track or a combined programming and math track. The math track covers probability and basic statistics at a level that’s somewhere between a quantitative finance interview and a GMAT quantitative section. Most software engineer candidates take the programming track. The questions are medium-difficulty DSA: you won’t see DP hard problems here, but a naive O(n^2) solution probably won’t pass the test cases on time.
Stage 2: HireVue
This is the stage that catches people off guard if they’ve never done a one-way video interview before. No interviewer. You record your answers to pre-set questions: typically five to eight questions, 30 seconds to read each, two minutes to answer. Two optional retakes per question. Goldman AI scores the responses, though humans do a final review before decisions are made.
Question types are behavioral plus a market awareness question or two. “Tell me about a time you disagreed with a technical decision and what you did about it.” “What’s a recent development in financial technology that interests you?” The behavioral questions are standard STAR format. The market question is the one people blank on because they didn’t think to prep for it.
One practical note: do your HireVue somewhere quiet with a clean background and stable internet. Goldman’s stated reason for rolling out HireVue was to “get to more places” and assess more candidates than campus visits allowed. The trade-off is that technical glitches make you look disorganized, even if the content of your answer is fine.
Stage 3: Technical phone screen (CoderPad, 60 min)
Live coding, usually CoderPad. One interviewer, sometimes two. The questions tend toward practical data structures and algorithms rather than competition programming. Goldman is a Java shop for many production systems, so if you know Java well, bring it. They’re not hostile to Python or Go, but OOP concepts come up more directly here than they do at pure web companies.
Finance-tinged problems appear here sometimes. “Given a stream of trade prices, find the maximum profit from a single buy-sell pair” is a sliding window problem dressed in trading clothes. The framing is intentional. Interviewers want to see whether domain context registers with you or whether it’s just noise. Say the words “bid-ask spread” correctly once and you’ve signaled more than a paragraph of explanation would.
Stage 4: SuperDay
Three to five back-to-back interviews, usually in a single afternoon. For campus hires it’s closer to three. For lateral mid-career roles it tends toward five. The mix for engineering tracks typically includes: one or two technical coding rounds, one system design round, one behavioral round, and a hiring manager conversation that is half behavioral and half “why Goldman.”
System design at Goldman is genuinely financial. Not “design Twitter” but “how would you architect a distributed market data feed that handles 100K price updates per second with guaranteed ordering?” Low-latency tradeoffs come up. Consistency vs. availability under network partition matters. If you’ve only prepped FAANG-style “design Instagram” problems, this is where you’ll feel the gap.
The “why Goldman over a tech company” question happens in almost every SuperDay. I’d argue it’s the single most predictable differentiator in the whole process. Candidates who have a specific answer (a real project they read about, a trading system architecture that interested them) pass. Candidates who give the “prestige and learning” answer in generic terms tend not to.
Goldman Sachs interview questions: what actually gets asked
These are patterns we see across the stages. Not a leaked question bank, just honest categorization.
Coding
- Implement an LRU cache, then explain how you’d adapt it for a trading system handling over a million requests per second.
- Design a rate limiter for an API. What happens when multiple servers share the same limit?
- Given a stream of stock prices, find the maximum profit from one buy-sell pair. Follow-up: at most two transactions.
- Implement a thread-safe singleton in Java. Walk through three different approaches and their tradeoffs.
- Design a time-series data structure optimized for fast range queries on financial tick data.
System design
- Design a real-time order book that handles equity market data from multiple exchanges.
- How would you build a fraud detection system that makes a go/no-go decision in under 100ms?
- Design a price aggregation service that normalizes market data from five exchanges with different schemas.
- How would you build a distributed cache for market data that degrades gracefully during a network partition?
The system design questions are harder than they read on paper because of the follow-ups. Saying “use Kafka for the message queue” is fine. Explaining how you’d handle message replay if a downstream consumer falls behind is what actually matters. Read about Goldman’s open-source quantitative finance work before your design rounds. It’s not directly testable, but you’ll think differently about the domain constraints.
Behavioral
- “Tell me about a time you delivered under significant time pressure. What corners did you cut, and which ones do you regret?”
- “Describe a project where your attention to detail prevented a serious problem.”
- “Why Goldman? Why not a pure tech company?” (Asked by almost every interviewer in the SuperDay.)
- “Tell me about a time you had to explain a technical decision to someone who didn’t have technical background.”
- “How do you stay current with what’s happening in financial markets and technology?”
The “attention to detail” theme appears constantly across Goldman interviews. Goldman posted a detailed interview preparation page on their careers site that frames this pretty directly. The behavioral competencies they list there map pretty tightly to what the SuperDay interviewers are grading against. Worth reading before you draft your STAR stories.
The tech teams worth understanding before your interview
Goldman engineering is not monolithic. Which team you’re targeting affects which questions you’ll get and which answers land well. There are four areas that come up most in the tech hiring we see:
Securities and Electronic Trading
High-frequency and algorithmic trading systems. Microsecond-level latency requirements. This is the team where knowing the difference between TCP and UDP for market data isn’t a trick question, it’s a real engineering decision you’ll make. C++ and Java, almost no Python in production. The technical bar here is the highest in the building.
Platform Engineering
Core infrastructure, cloud migration, developer tooling. More recognizable to someone coming from a traditional tech background. Goldman has been moving infrastructure to a hybrid cloud model over the past several years. If your background is platform or SRE, this is the entry point that makes the most sense.
Global Banking and Markets Technology
Client-facing systems, transaction processing, risk infrastructure. Heavy on Java. The “Post Trade Equities Enricher” engineering roles you’ll see on their careers page sit here. Less sexy name, genuinely interesting scale problems around settlement and reconciliation.
Risk and AI Engineering
Goldman’s head of AI engineering for asset management has said publicly that they’re now looking for three overlapping skill sets: AI engineering plus finance domain knowledge plus traditional software development. That’s a real signal about where hiring is going. If you have ML experience and any finance background, this division is worth targeting specifically.
Compensation: what the real numbers look like
Goldman pays well. The bonus culture means year-to-year variance is real, and “total comp” at Goldman can swing 20-30% in a down year in ways that stock-heavy Big Tech packages don’t. These figures are approximate and based on aggregated market data from sources like Glassdoor and Levels.fyi as of mid-2026; Goldman doesn’t publish salary bands.
| Level | Base | Bonus range |
|---|---|---|
| Analyst (New Grad) | $110K-$130K | $20K-$40K |
| Associate (3-5 yoe) | $150K-$180K | $50K-$100K |
| Vice President | $200K-$250K | $100K-$200K |
| Executive Director | $275K-$350K | $150K-$350K |
RSUs exist at VP and above. The bonus culture is real: Associate comp can land anywhere in a wide range depending on division performance and individual review. Don’t anchor your financial planning to the top of the range.
How to prepare, in order of what actually matters
This is the section where guides usually list six equal-weight bullet points. I don’t think they’re equal weight, so I’m not going to present them that way.
Most important by a meaningful margin:
Develop a real answer to “why Goldman over a tech company.” Not a polished answer, a real one. Read two or three of Goldman’s engineering blog posts on developer.gs.com. Find something specific that interests you. If the only honest answer is “better pay in finance right now,” that’s fine, but that’s not what you say in the interview. What you say is something specific you learned about their systems that you’d want to work on. The difference between candidates who pass SuperDay and those who don’t almost never comes down to algorithms. It usually comes down to this question.
Second:
Practice the HireVue format. Record yourself with no audience, 30-second read time, two-minute answer. It’s uncomfortable in a specific way that is different from live interviews and practice interviews with other people. The discomfort mostly goes away after a handful of dry runs. Goldman explicitly uses HireVue to reach more candidates than campus visits allowed, which means more people compete for spots than you might expect from an elite financial institution.
Third:
Prep your system design for finance-specific constraints. Low-latency tradeoffs. Ordered message delivery. Idempotency in transaction processing. This is where candidates coming from pure web backgrounds get surprised. If you’ve done our system design interview prep guide, you’ll have the right vocabulary. Goldman’s design questions add a constraint layer on top of those foundations.
On coding prep:
Medium LeetCode is sufficient for most Goldman tech roles. The HackerRank OA is medium difficulty. The phone screen is practical. The SuperDay coding round goes deeper, but it’s still not “hard” in the LeetCode sense. Grinding hard problems is a worse use of your prep time than understanding Goldman’s domain well. This is an honest opinion and probably a debatable one, but it’s what the pattern of sessions we see supports.
Want to see how your answers actually sound under pressure? Our AI interview copilot runs Goldman-style mock sessions and gives specific feedback on both content and delivery. It’s worth doing at least one full Goldman mock before your SuperDay, not because the questions will be identical, but because you’ll find out which STAR stories hold up under follow-up and which ones don’t.
Goldman versus the other finance-tech options
Goldman is not the only large financial institution that runs an interesting engineering shop. Jane Street and Citadel Securities are more intense on the technical bar and pay more. Two Sigma is heavier on ML and quant backgrounds. JPMorgan is larger and generally easier to get into but more variable on team quality. What Goldman offers specifically is brand weight, reasonably good engineering across multiple divisions, and a degree of operational stability that pure hedge funds don’t have. If you’re choosing between Goldman and a FAANG, the question comes down to whether you want finance context in your work. If you’re indifferent to that, take the FAANG offer. If the domain genuinely interests you, Goldman is a real peer, not a consolation prize.
For more context on how Goldman’s interview compares to other large tech companies, our FAANG interview questions breakdown covers the differences in format, depth, and typical timelines across Google, Meta, Amazon, and Apple. The behavioral prep in our behavioral interview guide translates directly to Goldman’s SuperDay format.
