Big Tech Cut New-Grad Hiring 65%. Find the Grads It Left on GitHub

If you are reopening junior engineering reqs, the strongest 2026 grads are more available than they have been in years. You just won't find them with a LinkedIn title search.

Indeed Hiring Lab's US software-development postings index hit 78.3 on October 2, 2026, up from 64.3 a year earlier. Over the same stretch, SignalFire's 2026 State of Talent report put new-grad hiring at the Tech Majors about 65% below 2019. Demand is coming back, and the biggest employer of junior engineers is still sitting out. A recent Ask HN thread on hiring junior engineers shows founders arguing over whether to hire them at all.

The software hiring rebound is a senior rebound

The recovery in software postings is real, but most of it is for senior engineers. Between May 2025 and May 2026, senior roles accounted for 71% of the increase in software-development postings, according to CNBC. Postings overall are still about 27.5% below their pre-pandemic level.

The junior share is worse. Indeed Hiring Lab found that entry-level roles made up just 4.5% of software-development postings in Q1 2026, the lowest share of any sector it tracks. Senior roles made up 69.3%, the highest.

4.5%

Why this matters for you: a strong grad with a real GitHub history is not fielding many offers right now. Few companies advertise for juniors, so grads who would once have been snapped up by a Tech Major are still open. The team that reaches out first, rather than posting a job and waiting, is competing against almost nobody.

Startups cut juniors harder than Big Tech did

Early-stage startups cut new-grad hiring even more than Big Tech did, so don't assume smaller companies have already absorbed the talent. SignalFire puts the drop at roughly 65% at the Tech Majors and about 76% at early-stage startups, both measured against 2019.

SignalFire defines the Tech Majors as Alphabet, Meta, Apple, Amazon, Microsoft, Netflix, Nvidia, Tesla, Uber, Airbnb, Block and Stripe.

The common story is "Big Tech left, so startups will scoop up the grads." The data says the opposite. Startups pulled back further, and one founder's reasoning in the Ask HN thread shows the mindset: "A senior with Max plan beats three junior hires (very rough estimate)." Read plainly, that is a bet that AI coding tools make one senior worth more than a junior team, so the junior req gets cut first.

That leaves an edge, but only for teams that actually reopen junior reqs. That group is small. Companies like MixRank (remote, global Junior Software Engineer posts asking for 0 to 3 years of experience plus "5+ years of curiosity"), Wildflower Health and Jefit have kept posting junior or junior-friendly roles in HN Who's Hiring through 2026. If you join them, you are in a short line.

Why LinkedIn search fails for junior engineers

Searching LinkedIn titles and headlines returns a huge, undifferentiated pool and misses the signals that separate strong grads from the rest. In Refolk's index of professional profiles, 421,877 US profiles carry an entry-level "Software Engineer" or "Software Developer" title. That number is the problem, not the answer.

A senior engineer's profile carries years of employers, titles and scope. A new grad's profile has a school, maybe an internship, and a generic headline. Thousands of them look identical. The differences that predict performance (what they built, who reviewed their code, whether anyone used it) live somewhere else.

The clearest example is Google Summer of Code. In GSoC, contributors spend a summer writing code for an established open-source organization under a mentor. In 2026, 184 mentoring organizations reviewed a record 23,371 proposals from 15,245 applicants in 131 countries and accepted 1,141 contributors. That is an acceptance rate of about 7.5%.

319

Only 319 entry-level or in-training profiles worldwide mention GSoC in their headline in Refolk's index. That is roughly 28% of a single year's cohort, and about 1.5% of the program's 22,000+ lifetime contributors. Measured against the 421,877 US entry-level profiles, at most about 0.08% (roughly 1 in 1,300) show it, and that is an upper bound, because the 319 is a global count. The seniority filters differ between sources, so treat these ratios as directional.

The mechanism is simple. Juniors don't write their profiles for keyword search. The strongest signal a new grad can have, selected at 7.5% and mentored by maintainers, rarely makes it into the one field recruiters search.

The sample also hints at who gets found. Among 25 sampled GSoC-headline juniors, Google (5) and Microsoft (3) were the top current employers, and 12 of 25 were in India. The ones who advertise the signal tend to get picked up. The ones who don't are your pool.

Which GitHub and open-web signals to search instead

To find junior developers on GitHub, search the artifacts their work leaves behind, not their self-descriptions. These are the four signals worth sourcing on, in rough order of how hard they are to fake:

1. Merged pull requests to real repositories. A merged PR means a maintainer who owes the author nothing reviewed the code and accepted it.
2. GSoC and similar mentorship program history. Selected at about 7.5%, with a public project page and a mentor on record.
3. Shipped side projects with real users. Deployed, linked, with issues filed by strangers.
4. Self-posted job-seeker activity. Contributions to new-grad job-list repos, which flag active availability.

Merged PRs to repos you recognize

Open-source history is the closest thing a new grad has to a work history. Look for merged PRs (not just opened ones) to projects with active maintainers. Weight a two-line bug fix merged into a large project above a sprawling personal repo nobody reviewed. Review comments are a bonus: you can read how the candidate handled pushback before you ever talk to them.

GSoC contributor archives, not GSoC headlines

GSoC publishes project pages, and the orgs keep the code. Start from 2026 mentoring orgs like the Apache Software Foundation, CNCF, the Python Software Foundation, KDE and OpenMRS. Work backward from their GSoC project lists to contributor GitHub handles, then to commit history after the summer ended. A contributor who kept committing after the program ended is more interesting than one who stopped the day it did.

Note that GSoC opened to non-students in 2022, so the alumni list is not limited to CS undergrads.

Shipped side projects

Small teams already screen this way. Deeter Analytics, hiring a junior ML engineer on HN in September, put it bluntly: "we hire on fundamentals and side projects you actually shipped." The word that matters is "shipped." Look for a live URL, a README with real install instructions, and issues or stars from people who aren't classmates.

Job-seeker repos on GitHub

SimplifyJobs/New-Grad-Positions has 18k stars and 348 contributors, and it carries a hacktoberfest tag, so its PR history is public. Class-of-2027 lists such as vanshb03/New-Grad-2027 and speedyapply/2027-AI-College-Jobs are already appearing. Many contributors to these repos are grads who are actively looking.

Treat this as a lead list, not a ranked list. Some contributors are maintainers or tooling authors, and adding a job link to a README says nothing about engineering skill. Use it to confirm availability, then rank on the other signals.

Grads who call themselves founders

SignalFire found that top CS grads in 2025 were twice as likely to call themselves a founder as the 2022 class, and 45% less likely to land a job at a Tech Major. "Founder" and "Co-Founder & CIO" show up inside Refolk's entry-level US sample. A search limited to "Software Engineer" titles misses every one of them. Some of these one-person companies are a shipped side project with a nicer title, and some of those founders would take a real job with a real team.

Stitching these sources together by hand is the slow part: GSoC pages, org repos, PR history, a personal site, a LinkedIn with three lines on it. That is the gap Refolk closes. You describe the grad in plain English and get back people matched across GitHub, LinkedIn and the open web, not a headline keyword match.

How to rank recent-grad engineers once you have a pool

Rank junior candidates on verified, reviewed work first and self-reported claims last. A simple scoring order keeps you honest when every profile reads "passionate about building scalable systems."

A few practical rules for applying it:

• Recency beats volume. Activity in recent months tells you more than a busy sophomore year.
• Persistence beats peaks. Commits after GSoC ended, or a side project with a second version, signal someone who keeps going without a deadline.
• Read one PR thread per finalist. How they responded to review predicts more than a take-home.
• Discount anything Tech-Major-adjacent you can't compete for. If they already have a Google internship conversion, move on. The point is the grads nobody else is calling.

This ranking step is where most junior sourcing stalls, because the evidence sits in different places for every candidate. Refolk returns the ranked shortlist with those signals pulled together, so the time goes into reading PR threads instead of opening tabs.

How to reach and vet them once you've found them

Reach out first, be specific about the work you found, and lower the risk on both sides with a paid trial. Because entry-level roles are only 4.5% of postings, a direct message that names a candidate's merged PR is one of very few a junior engineer will get.

A founder who opens with "I read your merged PR to KDE" is competing with almost nobody.

A few patterns worth copying from the Ask HN thread and recent Who's Hiring posts:

• Paid trial before full time. One founder in the thread said: "To vet junior hires I'm offering them paid internships (above median market pay) before making a FT hire decision." Above-market pay matters. It signals seriousness and keeps you from losing the candidate to the first full-time offer.
• Say junior plainly. With entry-level at 4.5% of postings, a post that names juniors as the target is rare. MixRank's "0 to 3 years" plus "curiosity" framing tells grads they are the target, not a stretch.
• Interview on their own code. Walk through the PR or project you found. It is a fairer test than a whiteboard, and it confirms they actually wrote it.
• Move in days, not weeks. The pool is open now because demand is senior-led. If junior postings recover, the grads with the best GitHub histories go first.

FAQ

Is entry-level software engineer hiring recovering in 2026?

Only slightly. Indeed's software-development postings index rose to 78.3 on October 2, 2026, from 64.3 a year earlier, but senior roles drove 71% of the increase. Entry-level roles were just 4.5% of software-development postings in Q1 2026, the lowest share of any sector Indeed tracks. SignalFire puts new-grad hiring about 65% below 2019 at the Tech Majors and about 76% below at early-stage startups.

How do I find junior developers on GitHub?

Search for evidence of reviewed work rather than profile keywords. Start from merged pull requests to maintained repos, GSoC project archives at orgs like Apache, CNCF, the Python Software Foundation, KDE and OpenMRS, and shipped side projects with outside users. Use contributors to job-list repos like SimplifyJobs/New-Grad-Positions as a list of people who are available, then rank them on the stronger signals.

What is the best way to vet a junior hire?

Look at work they have already done, then test it in a paid trial. Read one review thread on a merged PR to see how they handle feedback, interview them on their own project, and consider a paid internship at above-market pay before a full-time offer, which is what one founder in a recent Ask HN thread described doing.