Guide · Career
Two things decide this: applying early enough, and knowing which door you are legally allowed to walk through. Most international students get the second part wrong late, which is the expensive way to learn it.
This is not legal or immigration advice. Immigration rules, agency guidance, and processing timelines change, and how any of it applies depends on facts specific to your own status, school, and programme. Everything below describes how the categories work in general terms only. Verify your own situation with your school's Designated School Official (DSO) or a licensed immigration attorney before you rely on it, and never act on a guide — including this one — as if it were a determination about your case.
Summer internship hiring in software runs roughly a year ahead. Large firms open applications in late summer and early autumn for the following summer, and many close when the pipeline fills rather than on the posted date. Quantitative finance is earlier still, with some programmes opening and closing before the academic year is properly underway.
The single biggest error is treating spring as the season. By spring the large programmes are done. The second biggest is optimising the resume for months instead of submitting it — materials are rarely the bottleneck, submissions are.
This table assumes you are targeting the summer of year N and starting in March of year N minus 1. The hours column is a planning estimate, not a rule; the point is that the load is uneven and the heavy months are earlier than most students expect.
| Month | What the market is doing | Your job that month | Hours/week |
|---|---|---|---|
| Mar–Apr | Previous cycle finishing; nothing open for yours | Build the thing you will talk about. One shippable project, finished, deployed | 6–8 |
| May | Quiet | Rewrite the resume once, properly. Get the parse clean | 3 |
| Jun | Quiet; quant and trading firms start posting | Portfolio page live. LinkedIn rewritten. Start the target list | 4 |
| Jul | First large-firm postings appear, unannounced | Target list to 80 names, tiered. Set up the tracker. Apply to everything already open | 6 |
| Aug | Volume ramps hard; quant deadlines land | Apply daily. Referral asks go out. Begin interview pattern practice | 10–12 |
| Sep | Peak. Most large-firm reqs live; career fairs | Highest-volume month of the cycle. Fairs, referrals, online assessments | 12–15 |
| Oct | Peak continues; earliest reqs close as pipelines fill | Keep applying. First interview loops. Track which pattern you failed | 12 |
| Nov | Interviews dominate; second posting wave | Loops. Keep 5 applications a week going regardless | 10 |
| Dec | Quiet fortnight; decisions batch up | Finals, then a hard reset. Audit the funnel over the break | 4 |
| Jan | Second wave opens; mid-size firms post | Re-apply to the second wave. Refresh the list with mid-size targets | 8 |
| Feb | Startups and smaller companies hire | Where late applicants genuinely get hired. Direct founder outreach | 8 |
| Mar–Apr | Backfills, declined-offer replacements | Keep going. Offers do appear here, and they are real offers | 5 |
| May–Jun | Short-notice roles; research and lab positions | Fallback plan executes: research, TA, or a self-directed build | varies |
The single highest-leverage week in the whole table is the first week of September. Reqs are live, competition has not fully woken up, and career fairs put a human between you and the pile. Block it out in advance and treat it as a work week.
For an internship, the question is usually narrower than students assume. In general, F-1 students work internships under practical training authorisation obtained through their school or through USCIS — Curricular Practical Training, or pre-completion Optional Practical Training — rather than through a petition filed by the employer. That is a structurally different thing from the H-1B sponsorship question that attaches to full-time hiring.
The mechanics that determine eligibility — the one-academic-year requirement before CPT, how CPT is tied to a specific employer and specific dates, the 12-month OPT allowance and how part-time pre-completion OPT draws it down, the STEM extension — are laid out in the CPT versus OPT guide, and the enrolment requirement specifically in the one-year rule guide. Neither of those is a determination about you either. Your DSO is the only person who can authorise CPT, and an attorney is the only person who should be telling you what your options are.
What you can act on without any of that being settled: the authorisation timeline runs on its own clock, and it is longer than the hiring timeline. Whatever route applies to you has processing time attached, and that processing time has to be finished before a start date in May or June. Working backwards from an offer you have not received yet feels absurd in September and is the reason people are ready in April.
Nearly every US application form asks two separate questions that students routinely collapse into one. They are asking different things, they feed different filters, and answering them inconsistently across applications is how a real offer gets withdrawn at background-check stage.
| The question as written | What it is really asking | How students commonly approach it | Why |
|---|---|---|---|
| Are you legally authorised to work in the United States? | Will you hold valid work authorisation for the dates of this role, without us filing anything? | Answered on the basis of the authorisation they will actually hold for the internship dates, confirmed in advance with a DSO | It is a factual question about your status, not a prediction. Confirm the fact before you answer it |
| Will you now or in the future require sponsorship for employment visa status? | Will we eventually have to file a petition to keep employing you? | For many international students the honest answer for future full-time employment is yes | Some employers filter on this regardless of the internship. You cannot argue past that filter, and misstating it is far more expensive than being filtered |
| Are you a US citizen or lawful permanent resident? | Eligibility for export-controlled, cleared, or federal-contract work | Answered factually; a no is disqualifying for those specific roles only | This is not the same filter as the sponsorship question. Many commercial roles never ask it |
| Free-text: describe your work authorisation | The recruiter wants one sentence they can paste to their immigration team | One plain sentence naming the status and the mechanism, no essay | Ambiguity here creates a phone call. A clear sentence creates a next step |
Two rules that hold regardless of your situation. First, answer accurately every time and identically every time — these answers are stored, compared, and re-asked at offer stage. Second, do not let a recruiter tell you what your authorisation is. Recruiters are not immigration advisers, they are frequently wrong in both directions, and a confident wrong answer from a company is still your problem. Take it to your DSO.
This is knowable, and almost nobody does it. Spending forty minutes here changes where the next eighty applications go, which is worth more than another week of resume edits.
| Source | What it tells you | What it does not tell you |
|---|---|---|
| US Department of Labor disclosure data (LCA and PERM filings) | Which employers have filed for foreign workers, in what job titles, at what wage levels, in which years | Anything about internships, which usually do not appear at all; it is a lagging full-time indicator |
| USCIS H-1B employer data hub | Approval and denial counts by employer and fiscal year | Whether they still do it, or whether the policy has changed under a new head of recruiting |
| The req text itself | The hard filter. Phrases like 'will not sponsor now or in the future' or 'must be authorised to work without sponsorship' are the answer | Whether an unstated policy exists. Silence is genuinely ambiguous |
| Alumni search on LinkedIn, filtered by your school and the company | Whether people whose earlier education was outside the US have interned there recently | Their status, which you should not ask about directly. Read it as a weak positive signal only |
| International student groups and upperclassmen | The most accurate signal available: who actually got authorised there last year | Anything about companies nobody in your circle has applied to |
| Asking the recruiter at a career fair | The current policy, from the person who administers it | A binding commitment. Get the answer, do not treat it as a promise |
There is a fuller walkthrough of the verification process, including exactly what to ask and how to read an ambiguous req, in the sponsorship verification guide.
A target list is not a list of companies you like. It is a portfolio with a deliberate risk profile, and the tiers exist so that a bad month in one tier does not end the cycle.
| Tier | Definition | Count | Applications | What it is for |
|---|---|---|---|---|
| Tier 1 — explicit | The req or the careers page states sponsorship is available | 8–15 | All of them | Highest conversion per application. Apply the day they open |
| Tier 2 — documented history | Public filing data plus alumni evidence of past hiring | 20–30 | All of them | The core of the list. Where most offers come from |
| Tier 3 — silent | No statement either way; usually mid-size product companies | 30–50 | Volume | Cheap lottery tickets. Genuinely ambiguous, worth the click |
| Tier 4 — explicit no | Req requires citizenship, permanent residency, or clearance | any number | Zero | Not pessimism, arithmetic. These applications cannot convert |
| Tier 5 — non-company | University labs, research groups, TA and grader roles | 5–10 | All of them | The fallback that is not a waste of a summer |
Illustrative planning arithmetic, not measured rates: if you send 100 applications and 12 produce any human response, and half of those reach a first round, and a third of those reach a final, you are looking at roughly one to two finals. That funnel is why the answer to a quiet October is more applications rather than a better resume. Run your own numbers once you have your own data — the point of the tracker below is that after eight weeks you stop guessing.
One referral outperforms a large number of cold applications, because it flags your record for human review rather than leaving it in a search result. The reason most referral asks fail is that they ask for a favour with no defined edges. Ask a yes-or-no question about a specific req instead.
SCRIPT 1 — cold alum, same school, no prior contact Subject: SJSU CS 2028 - quick question about req 12345 Hi [name] - I am a CS undergrad at SJSU, graduating 2028. I saw you are on the platform team at [company]. I applied to the SWE intern req 12345 last week. Would you be comfortable submitting a referral for it? Completely fine if not - I know referrals are personal. For context: [one line, the single most relevant thing you built, with a link]. Thanks either way. SCRIPT 2 — classmate or friend already at the company Hey - are you able to refer for req 12345 (SWE Intern, Platform)? Happy to send you my resume and a two-line blurb so it takes you thirty seconds. If your referral quota is already spent this cycle, no problem at all - just say so. SCRIPT 3 — recruiter you met at a career fair, sent within 24 hours Subject: Followed up as promised - [your name], SJSU, STEM fair Tue Hi [name] - we spoke on Tuesday at the SJSU STEM fair about the platform intern role. You mentioned [specific thing they said]. I applied to req 12345 that evening. Resume attached, and the project I described is live at [url]. One question I wanted to confirm rather than assume: does that programme consider candidates who will be working on practical training authorisation obtained through their school? Thanks for the time.Three rules underneath all three scripts. Include the req number, because a referral form asks for one and hunting for it is the friction that kills the favour. Give an explicit exit, because an ask with a graceful no attached gets answered. And send it within a day of applying, so the referral and the application land together.
Applications die in the gaps between them. The gaps are always the same three: you did not record the req number, you did not record the deadline type, and you did not record the next action with a date on it.
company,role,req_id,tier,source,url,deadline_type,deadline_date, applied_date,status,next_action,next_action_date,referral, sponsorship_signal,notes deadline_type : hard | rolling | not_yet_open status : found | applied | oa_sent | oa_done | phone | onsite | offer | rejected | ghosted sponsorship_signal : explicit_yes | history | silent | explicit_no| Field | Why it exists | What breaks without it |
|---|---|---|
| req_id | Referral forms and follow-ups both require it | You cannot ask for a referral without re-finding the posting |
| deadline_type | Separates hard dates from rolling reqs | You treat a rolling req as safe and it closes when the pipeline fills |
| applied_date | Drives the follow-up clock | You follow up at day 3 or never |
| next_action_date | The only field that makes it a system rather than a log | Everything sits at status applied forever |
| sponsorship_signal | Lets you measure your own conversion by tier | You never learn whether tier 3 is worth the clicks |
| status | Funnel analysis at the end of the cycle | You cannot tell whether the problem is the resume or the interviews |
A spreadsheet is fine. Apply OS is the free browser-only version of exactly this schema if you would rather not build it, and it keeps the deadline radar in front of you instead of in a tab you closed.
| Company type | Loop shape | Typical elapsed time | What it actually tests |
|---|---|---|---|
| Large tech | Online assessment, then 1–2 technical rounds, sometimes a team or host match | 4–10 weeks | Pattern recognition under time pressure, then communication while coding |
| Quant and trading | Timed OA heavy on probability, then several fast technical rounds | 1–3 weeks, sometimes days | Mental maths, probability, precision. Speed is part of the grade |
| Mid-size product | Recruiter screen, one technical, one team fit | 2–4 weeks | Whether you can ship. More practical, less puzzle |
| Startup | Founder call, then a take-home or a pairing session, then team | 1–2 weeks | Whether you build things without supervision |
| Research lab or university group | Conversation about a paper or a project, no algorithm round | Weeks to months | Whether you can read, reason, and finish something hard |
| Consulting or strategy | Behavioural, then case interviews | 3–6 weeks | Structured reasoning out loud, arithmetic without a calculator |
Practice should match the shape you are actually facing. Grinding random problems is the least efficient version of this; tracking which pattern you failed is the efficient version, and the pattern inventory is in the interview patterns guide. If quant is on your list, the calendar is materially earlier and is covered separately in the quant internship timeline.
Assume for a moment that it is February and you have nothing. This is common, it is not evidence about you, and there is a diagnostic order to work through rather than a general feeling to have.
The process is high-volume and mostly silent. Rejection rates are brutal for everyone and worse when a sponsorship filter is running against you. The only counter is throughput plus a small number of high-quality signals — the portfolio, the research, the referral — that make you unusual in a stack of similar resumes.
The part that is genuinely different for international students is not difficulty, it is sequencing. Everyone else can accept an offer and figure out logistics afterwards. You cannot, because the authorisation timeline starts before the offer does. That is the whole disadvantage, and it is manageable precisely because it is a scheduling problem rather than a mystery.
Everything I built for my own search is public: Apply OS for the pipeline and deadlines, Visa Navigator for laying out the authorisation timeline alongside it, and the resume rebuild if you would rather hand the document off. None of it is advice about your case, and all of it assumes you are talking to your DSO in parallel.
FAQ
Roughly a year ahead. Large firms open applications in late summer and early autumn for the following summer and often close early when the pipeline fills. Quantitative finance runs earlier still. Spring is when smaller companies and startups hire, not when the large programmes are available.
Usually not in the way full-time roles do. In general, F-1 students work internships under practical training authorisation obtained through their school or USCIS rather than through a petition the employer files. Full-time hiring is where the sponsorship question becomes an employer cost. Confirm what applies to you with your DSO or an immigration attorney.
Answer both questions accurately, separately, and identically across every application, since the answers are stored and re-checked at offer stage. Confirm the factual position with your DSO before you answer rather than relying on what a recruiter tells you, because recruiters are not immigration advisers and are frequently wrong in both directions.
Check public Department of Labor disclosure data and the USCIS employer data hub for filing history, read the requisition text for explicit statements, search LinkedIn for alumni from your school who studied outside the US and interned there, and ask recruiters directly at career fairs. Filing data is a lagging full-time indicator and rarely covers internships.
Companies that state sponsorship availability outright, companies with documented filing history plus alumni evidence, mid-size product companies, startups with technical founders, and university-adjacent research roles. Requisitions that require citizenship, permanent residency, or a security clearance cannot convert and are not worth an application.
Referrals, shipped projects with live links, and any research output. A referral flags your record for human review rather than leaving it in a search result, and a clickable project beats a bullet describing one. Materials are rarely the bottleneck; submission volume usually is.
Generally yes, because some employers filter on the sponsorship question regardless of the internship's actual requirements. The counter is applying earlier, applying in volume, and carrying one or two unusual signals such as research or a shipped product.
Diagnose which stage is failing, because no responses, failed assessments, and failed final rounds have entirely different fixes. Then work the February and March wave, approach small companies directly, consider a university research or lab position, and treat a finished shipped project as a legitimate outcome rather than a consolation prize. Do not take unauthorised work to fill the gap.