The Tech Internship Interview Guide

The internship loop is not the new-grad loop with the difficulty turned down. The format is genuinely different. The OA bar at the intern level is the gate, not the technical interviews. The recruiters move on a different timeline. And the conversion rates - whether the internship turns into a return offer - vary by 4-5x between top programs.

Most college students prep wrong. They watch new-grad interview YouTube, do hard Leetcode for three months, and then bomb the OA because nobody told them that 4 mediums in 70 minutes is the actual bar.

Here is the manual you actually need.

Timeline: Apply Earlier Than You Think

The single most common mistake is applying too late. Big tech internship recruiting opens in August-September for the following summer. By December, the top programs have closed.

Rough recruiting calendar for Summer 2026 internships (the typical pattern; specific companies vary by 2-4 weeks):

August 2025: Google STEP, Meta U, Microsoft Explore (the "early-career" diversity-focused programs) open.
September 2025: Google general SWE intern, Meta SWE intern, Amazon SDE intern, Apple SWE intern open. Most large hedge funds and trading firms (Jane Street, Citadel, HRT, Two Sigma) recruit hard in this window.
October-November 2025: OAs for the September applications go out. First-round interviews start.
December 2025-January 2026: Most FAANG offers go out. Many top programs are full by end of January.
February-April 2026: Mid-size and Series A-D startups recruit. Some FAANG teams that did not fill their quota in fall reopen.
May 2026: Late-stage scramble. Smaller companies, late budget approvals.

If you are reading this in May and trying to start prep for "this summer," you are mostly out of options at the top tier. Lock in next August.

The Loop: Shorter, More OA-Heavy

The standard FTE loop is 5 onsite rounds. Internship loops are usually 1-3.

Typical FAANG internship loop:

Online Assessment (OA). 60-90 minutes, 2-4 medium Leetcode problems. Auto-graded. This is the gate. ~70-80% of applicants are filtered here.
One technical interview. 45 minutes. One coding problem (usually one medium, sometimes a medium + an easy follow-up), some quick discussion of why you want to work there. Sometimes a brief behavioral.
Sometimes a second technical interview for the more competitive teams. Same shape - one coding problem, some chat.

That is it. Three contact points: OA, interview 1, interview 2. No system design (intern-level loops do not include SD; you are not expected to know it). Minimal behavioral (one or two questions, low weight). No bar-raiser. No team match for most companies (you are matched to a team after the offer for FAANG; matched before the offer for some startups and trading firms).

The loop is short because the company is committing to ~12 weeks of you, not 4 years. The risk is lower; the bar is lower.

Compare to the new-grad / FTE loop:

4-5 onsite rounds (1-2 coding, 1 system design, 1 behavioral, sometimes 1 domain)
Bar-raiser interview (Amazon)
Team-matching process before / after offer
Multi-week debrief and committee process

For the same student in the same year, an internship loop takes ~3 weeks total; the FTE loop takes ~6-10.

What's Actually Asked at the Intern Level

The technical bar is real but the topics are narrow.

Heavy DSA, light system design.

You will be asked Leetcode-style algorithm problems. The level is "medium" and the most common topics:

Arrays / strings (~30% of intern questions)
Hash maps and sets (~20%)
Trees - especially binary trees (~15%)
Two pointers / sliding window (~10%)
BFS / DFS on graphs or grids (~10%)
Sorting and basic searching (~5%)
Stacks / queues (~5%)
Easy DP - mostly 1D (~5%)

You will not typically be asked:

Hard DP (no longest-common-subsequence variants, no interval DP)
Advanced graph algorithms (no Dijkstra, no MST, no Tarjan)
Segment trees, Fenwick trees, suffix arrays
Concurrency
System design

Drilling Hards is wasted prep at the intern level. The hard problem will not appear. Drill 100-150 mediums on the topics above instead.

Behavioral is light but real.

Intern interviews include 1-2 behavioral questions, usually:

"Why this company?"
"Tell me about a project you worked on."
"Tell me about a time you worked with a difficult teammate." (Less common than at FTE level.)

The bar is "you can hold a conversation, you can articulate what you did, you do not seem like a jerk." STAR format helps but is not strictly required - intern interviewers grade much more loosely on behavioral than FTE interviewers do.

Side projects matter more than at FTE.

For interns, the resume is usually thin (one or two prior internships or none, course projects, maybe a hackathon). Side projects fill the gap and they actually get read.

What works:

Something you built end to end. Not a tutorial follow-along; something you decided to build and shipped. Bonus if it has real users (even five users counts).
Something with depth. A Discord bot is a starter project; a Discord bot that uses a custom message-queue layer because the standard library was too slow for your use case is a portfolio piece.
Something deployable and demoable. A live URL, a video walkthrough, or a working binary the interviewer could run.

What does not work:

"I built a TODO app in React." (Everyone has built a TODO app.)
"I cloned [popular product]." (Reads as exercise, not interest.)
"I worked through 30% of a course on Coursera." (Not a project.)

OA Platforms: Know the Format

Most internship OAs run on one of three platforms, and the quirks of each platform matter.

HackerRank. Used by Amazon, IBM, Goldman Sachs, JPMorgan, many others. Standard Leetcode-style problems with their own editor. Usually 60-90 minutes, 2-4 problems. The custom test case feature lets you write your own test inputs - use it; the provided test cases are not exhaustive and you will need your own.

CodeSignal General Coding Assessment (GCA). Used by Meta (formerly), Roblox, Quora, others. The GCA is a standardized 70-minute assessment with 4 problems of escalating difficulty (easy → easy/medium → medium → medium/hard). You get one score (0-850 range; ~750 is FAANG-tier). This score can be reused across companies that accept it - take it once, well, and apply broadly.

CodeSignal Industry Coding Framework (ICF) / company-specific. Some companies write their own assessments on CodeSignal. These look like standard OAs but are bespoke - 1-3 problems, 60-90 minutes, often with company-specific themes.

Hackerrank, Codility, or company-built proprietary OAs. Various: Bloomberg, two-sigma, Citadel, etc. all have their own. Format varies.

Google's intern OA. Custom. Usually 90 minutes, 2 problems, run on Google's own platform. Higher difficulty than the average FAANG intern OA - both problems are usually solid mediums. Strong signal at this stage; weak performance here often gates you out.

Universal OA tactics:

Read all the problems first, solve the easiest one first, save the hardest for last.
A correct, clean easy/medium beats a half-implemented hard. Partial credit is rare.
Test edge cases - empty input, single element, max-size input, duplicate values.
If you finish early (you usually will not), use the time to refactor and add comments. Some OAs have human review of high-scoring submissions.
Practice in the actual editor before the test. If you have always coded in VS Code with autocomplete, the bare-bones HackerRank editor will surprise you.

Prep on a College Schedule

You are taking a full course load. You do not have 8 hours a day to grind Leetcode. The good news: you do not need them.

Realistic prep plan, starting July before fall recruiting:

July (before applications open):
- Resume polish. Two specific projects, with metrics. One past internship or course project as the headliner.
- GitHub cleanup. Top 3 pinned repos, all with READMEs.
- 30-50 Leetcode easies + mediums to warm up. Topics: arrays, hash maps, two pointers, basic trees.
August:
- Apply early - first week, every company that has opened.
- Continue Leetcode: 5-7 problems/week, focus on patterns from the topics above. Target: 100 mediums by November.
September-October:
- Apply broader - everything that opens.
- First OAs land. Knock them out within 48 hours of receiving them.
- Mock OA: do a full timed 90-minute, 4-problem set every other week. The timing skill is a separate skill from solving.
November-December:
- Real interviews land. Do 2-3 mock interviews with friends, classmates, or career-services staff before your first real one.
- Prep your "tell me about your project" answer. 2 minutes, with one technical detail you can defend, one tradeoff you made.
- Behavioral prep: 4-5 short stories that cover collaboration, technical challenge, and the "why this company" question.
January-March (if still searching):
- Continue applications - mid-size companies recruit later.
- Up your problem count if you have time. Switch focus to medium-hard if you are weak there.

The total real time investment: ~6-8 hours/week for 4 months. That is a manageable pace alongside coursework if you start early enough.

If you start in November for next summer, you are still fine - tighten the timeline, drop the warm-up phase, do 8-10 hours/week.

Top Internship Programs Worth Applying To

Not all internships are equal. Some convert at 80%+ to full-time offers; some are essentially marketing programs for the company brand. Here is a rough tier list:

Tier 1 (top programs, FAANG and equivalent):

Google STEP - Sophomore-focused early-career program. ~3% acceptance rate, ~70% conversion to a Google FTE offer.
Google SWE Intern (general) - Junior/senior. ~75-85% conversion.
Meta U - Sophomore-focused, similar to STEP. ~3% acceptance, ~70% conversion to Meta FTE or to the regular Meta intern program.
Meta SWE Intern (general) - Junior/senior. ~80% conversion historically; lower in 2023-2024 due to layoffs, recovering in 2025-2026.
Microsoft Explore - Underclassman (freshman/sophomore) program. Less coding-intensive than the standard intern; more rotational. ~50% conversion to standard Microsoft intern next year.
Microsoft SWE Intern - Junior/senior. ~70-75% conversion.
Amazon SDE Intern - Junior/senior. Highest volume of intern offers in tech (~3000/year). ~50-60% conversion - lower than Google/Meta because Amazon hires more interns.
Apple SWE Intern - Junior/senior, often team-specific. ~60-70% conversion.
NVIDIA SWE Intern - Highly competitive in 2024-2026 due to AI demand. Conversion rate ~70%.

Tier 1.5 (top tech, often paying as well or better than FAANG):

Stripe Intern - Strong technical bar, structured intern program. ~70% conversion.
Databricks Intern - High bar. ~60-70% conversion.
Citadel SWE Intern - Highest pay in the industry ($14-18K/month). Brutal interview bar - more like quant FTE than intern. ~75% conversion if you accept the offer.
Jane Street Intern - Comparable pay and bar. ~80%+ conversion.
Hudson River Trading, Two Sigma, Optiver - Same tier.

Tier 2 (good programs, lower conversion or smaller scale):

Anthropic, OpenAI - Newer intern programs, very high bar, conversion uncertain at scale.
Snowflake, Cloudflare, Datadog, MongoDB - Strong programs, ~50-65% conversion.
Salesforce, Adobe, Workday - Larger volume, ~50% conversion.

Tier 3 (apply if you do not get tier 1/2):

Most large enterprises (banks, insurance, healthcare) - paying $40-70/hr, conversion uncertain, the experience is generally less rigorous than tier 1.
Series A-C startups - highly variable. Some are amazing; some are unstructured and you spend the summer wondering what you should be doing. Vet hard - ask about the specific projects you would work on, who you would report to, and how previous interns spent their summers.

Numbers to anchor on: A FAANG SWE intern in 2026 makes $11-14K/month. Trading firms hit $14-18K. Top startups hit $9-13K. Mid-tier $7-9K. Below $5K/month, the only reason to take it is the experience filling a resume gap.

The Return-Offer Game

The return offer is the actual goal of an internship - it is FTE recruiting six months early, with a 70-90% acceptance rate at top companies vs ~5-10% for the general FTE applicant pool.

To maximize return-offer odds:

Pick the right project, day one. Most interns get assigned a project; some have flexibility. If you have flexibility, pick something that has a clear, demoable outcome by week 10. Vague research projects, infrastructure work without a customer, or "explore X" projects are riskier - they do not produce a clean story for your end-of-internship presentation.

Ask for feedback at week 4. Mid-internship review. Most managers will give vague positive feedback unless you push. "Specifically, what would I need to do over the next 6 weeks to get a strong return-offer recommendation?" gets you concrete answers. Apply them.

Ship a measurable thing by week 10. The end-of-internship presentation is a major signal. "I built X, deployed it to production, here are the metrics" is a winning story. "I researched some approaches to X" is not.

Network laterally, not just vertically. Your manager writes the return-offer recommendation, but your peers and other engineers on the team weigh in. Eat lunch with people. Pair with engineers who are not your manager. The signal "everyone on the team likes working with this intern" matters more than people think.

Do not coast week 11-12. A surprising number of interns mentally check out the last two weeks. The ones who do not - who keep shipping, keep documenting, keep showing up - get the offer.

The return offer is also negotiable. Companies will give you "exploding" return offers (deadline 4-6 weeks after the internship ends) to lock you in before you finish your senior year. You can usually push for a 10-20% bump if you have a competing offer from another internship's return offer.

Common Mistakes

Applying too late. Already covered. Worth repeating.

Spamming 200 applications without tailoring any of them. Better to send 50 well-targeted applications (with a real cover line for each) than 200 form applications.

Treating the OA as a warmup. The OA is the gate. Most rejections happen there. Take the first OA you get as seriously as you would take an onsite.

Over-prepping for hards. Wasted time at intern level. Mediums are the bar.

Ignoring the behavioral. Even though it is light, a bad behavioral answer can flip a maybe to a no. Prep 4-5 short stories.

Picking a brand-name internship over a learning-rich one. A summer at Google doing meaningless infrastructure tickets is worse for your career than a summer at a 50-person startup shipping a product feature you can demo. (Caveat: the brand opens doors for the next round of recruiting, so weight accordingly.)

Not asking about the team / project before accepting. A FAANG offer at the wrong team can be a worse summer than a tier-2 offer at a great team. Ask: who would I report to? What would my project be? Can I talk to a previous intern from that team?

What Actually Matters in Year 1 of Recruiting

If you are reading this as a freshman or sophomore: the most important thing you can do this year is not Leetcode. It is building one substantial side project and getting your first internship anywhere.

The first internship is the hardest. Everyone wants someone with experience; you have none. Take the smaller, less-prestigious offer in your sophomore year. Use it as a stepping stone. The transition from "no internship" to "any internship" is the single biggest difficulty jump in early-career recruiting; the transition from "any internship" to "FAANG internship" is much smaller.

By junior year, with one internship under your belt and one shipped side project, your application reads completely differently from the same student a year earlier. That is where the FAANG offers come from.

Start early. Apply broad. Take the OA seriously. The rest follows.

gitGood drills the medium-tier coding patterns that show up on intern OAs - sliding window, two pointers, BFS/DFS, hash-map tricks, easy DP. Free tier covers 5 free question claims; premium unlocks all 1000+. $5/month, no fluff.