The New Grad Software Engineer Interview: 12 Weeks From Zero to Onsite

The funnel: where new grads actually get cut

Before the plan, understand where the cuts happen. The new grad SWE funnel at most large companies looks like this: 1000 resumes → 300 OAs → 80 phone screens → 25 onsites → 5 offers. Each transition has a specific failure mode and a specific way to defeat it. Knowing where you actually are in the funnel is the first move.

The math says the largest absolute number of candidates die at the resume stage, but the largest percentage of qualified candidates die at the OA. That is the highest-leverage stage to target if you have any technical ability — it is fixable in a few focused weeks.

Weeks 1-2: resume + project polish

Resumes are scanned for six seconds. Make every line earn its placement. Three things matter: a clear top section (school, GPA if 3.5+, target role, contact), one or two strong projects that demonstrate independent thought, and any internship experience expressed in impact rather than tasks. Drop coursework lists, soft skills, and anything that looks like a CS101 assignment.

• Lead each project bullet with a verb and end with a measurable outcome. 'Built a recommendation system that increased click-through by 12% on a 5,000-user user base' beats 'Worked on machine learning project.'
• Pin one project that you can talk about for 20 minutes — it will come up in every interview.
• Cap the resume at one page. Two pages signals you do not know what to cut, which signals immaturity.
• Get three peers to read it cold for 30 seconds and tell you what they remember. If they cannot remember any project, the projects are weak or the resume is cluttered.

Weeks 3-6: LeetCode patterns, OA preparation

Online assessments are the part most new grads under-prepare and the part where the biggest absolute number of candidates die. Most OAs are 70 minutes, two medium problems, hidden test cases. The pattern coverage is narrower than people fear — about eight patterns cover most OAs.

1. 1Two pointers and sliding window (high frequency).
2. 2BFS/DFS on grids and trees.
3. 3Stack and queue (valid parens, monotonic stack basics).
4. 4Hash map counting (anagrams, frequency questions).
5. 5Heap top-K and merge-K.
6. 6Basic dynamic programming (1-D, climbing stairs and friends).
7. 7Sorting + interval problems.
8. 8Binary search (classic, plus search-the-answer-space).

Drill three problems per pattern per week, time-boxed to 25 minutes per problem. Review every solution against the optimal one. By week six you should solve a fresh medium in under 20 minutes 70% of the time. That target is calibrated — it is what an OA pace looks like under stress.

Weeks 7-9: behavioral and system-design-lite

New grad loops increasingly include a behavioral round, even at companies that historically skipped it. Some also include a 'system design lite' round — design a URL shortener, a chat app, a simple cache. Both are graded with new-grad calibration; you do not need senior depth, but you need structure.

Build a story bank of six stories: a project you led on a team, a time you debugged something hard, a time you disagreed with a teammate, a time you learned something fast, a time you missed a deadline and recovered, a time you helped someone else. Use the SCOPE-AID structure but compressed — your stories will be 2-3 minutes, not 4.

For system design lite, learn one canonical answer for each of: design a URL shortener, design a chat application, design Twitter (basic), design a rate limiter. Practice them on the whiteboard. The interviewer is looking for whether you can name components (load balancer, API server, database, cache) and walk one request end to end. They are not looking for production-grade tradeoffs at this stage.

Weeks 10-11: mocks and full-loop simulation

By week ten you should be doing two mock interviews per week with a peer or paid coach. The point is calibration and stamina. The first time you do four interviews in one day, you discover that mental fatigue affects your fifth-round performance more than your skill does. Simulating that day in advance is the only way to prepare for it.

1. 1Week 10: three mocks across coding, behavioral, and system-design-lite. Recorded if possible.
2. 2Week 11: one full simulated loop day (4 rounds in 5 hours). Debrief same evening.
3. 3After each mock, write the post-mortem template. Identify one specific behavior to change before the next session.

Week 12 and after: applying, recruiter calls, closing

Apply in waves, not in a single avalanche. Submit to your top ten target companies week twelve, your next twenty week thirteen. Follow up on every silent application after two weeks with a one-line email to the recruiter (find them on LinkedIn). Personal referrals are the highest-conversion application path; ask second-degree connections for them, with a short blurb they can paste.

When recruiter calls start coming in, be ready with a 30-second self-pitch (school, focus, one project, one accomplishment, target role) and three thoughtful questions about the team and the work. Recruiter calls are graded — they decide whether you advance to the technical screen and what 'tier' of role you are routed toward.

International students and bootcamp graduates: what changes

The default new grad playbook assumes a US citizen or permanent resident with a CS degree from a known university. Two large groups of candidates fall outside that assumption — international students requiring sponsorship, and bootcamp graduates without traditional degrees — and the tactics shift meaningfully for each.

International students should target sponsorship-friendly companies first. Maintain a list (updated each cycle on Blind, Reddit, or company career pages) of who actually sponsors at the new grad level. Apply broadly to those companies; deprioritize companies that historically do not. Lead with technical strength in the resume — H1B reform conversations make sponsorship cost-sensitive, and a candidate who clearly clears the bar is easier to sponsor than one who is on the fence.

Bootcamp graduates face a sharper resume cut. Counter it with two things: a portfolio of two or three substantial projects (each over a thousand commits, with real users if possible), and any open-source contribution to a recognized library. Both are visible signals that survive an automated keyword filter that strips on the missing degree. Personal referrals are also disproportionately valuable here — they shortcut the resume cut entirely.

Both groups face the same advice on the bigger arc: the first job is harder than the second. Once you have one US tech role on the resume, the next round of recruiting is dramatically easier. Optimize your first job for learning velocity and the brand on the resume, even if the comp is not at the top of band. Year-two compensation on the second offer will more than make up the difference.