Skills-based hiring for software engineers means you decide on evidence of what someone can do (code they wrote, problems they solved in front of you) instead of proxies like a degree, a famous employer or years of experience. Everyone says they do it. Very few do. When Harvard Business School and the Burning Glass Institute studied companies that dropped degree requirements, the change showed up in not even 1 in 700 hires.
So the hard part is practice, not belief. Here is what it looks like when an engineering team actually does it.
Key Takeaways
- Removing "BS in Computer Science required" from a job post is the easy part. The hard part is changing how you screen and interview.
- LinkedIn's research found a skills-based approach makes the US talent pool 15.9 times larger, the highest of 56 countries studied.
- For engineers, the skills test has to include working with AI tools. Testing without them measures an older job.
- Proof travels better than claims. A verified assessment result is worth more than a line on a resume.
Why Most Skills-Based Hiring Stays on Paper
The same Harvard and Burning Glass report found that about 45% of firms that announced the change made it "in name only". The job post changed. The resume screen didn't.
That tracks with what we see. A founder removes the degree line, then the recruiter (or the founder at 11pm) still skims for Stanford, Google and "8+ years". The proxies come back because they are fast. Reading a GitHub repo takes 20 minutes; reading a logo takes two seconds.
The fix is to make the fast path skills-based too. That means a short, standard test early in the funnel, scored the same way for everyone, before anyone looks at pedigree.
Skills-Based Hiring for Engineers, Step by Step
1. Write the job around outcomes. "Own the billing service and cut failed payments" beats "5+ years of Java". A hiring brief forces this.
2. Name 4 to 6 skills, and how you will see each one. Data modeling: a schema exercise. Debugging: a broken test in a real repo. Communication: a written design note.
3. Test before you screen resumes, or at least in parallel. A 25 to 45 minute work sample, same for every candidate.
4. Allow AI tools in the test, and watch how they use them. The coding test vendors have moved this way: HackerRank lets candidates pick Claude, Gemini or GPT inside the test, and CodeSignal sells agentic assessments with Claude, Cursor and Codex. Someone who pastes output without reading it is a different hire from someone who checks it.
5. Score with anchors. Use an interview scorecard with written descriptions for each score, so two interviewers rate the same behavior the same way.
6. Look at pedigree last, if at all. By then you already have evidence.
LinkedIn's 2025 Future of Recruiting report found that 72% of recruiting professionals say improving how skills are assessed will be a priority over the next 12 to 18 months. Priority is cheap. Steps 3 and 5 are where it gets real.
How Ruzora Applies It
Our bench is skills-based by construction. Nobody gets on it because of a school or an employer. Every engineer clears a graded coding assessment (70 out of 100 or better) and an AI-led technical interview (3.0 out of 5 or better). About 1 in 5 applicants pass both. Blind profiles, which Sol shows in the chat on our homepage when your stack is on the bench, show no name and no employer, so you look at the evidence before you look at anything else.
Engineers who pass can carry a "Ruzora Vetted" badge with its own public verification page, and add it to LinkedIn as a certification that links back to that page. It is a small thing, and we have only started issuing them, but it points at the right idea: a skill claim that anyone can check. Making these credentials machine-verifiable is on our roadmap.
A Concrete Version
Picture a 40-person logistics startup that needs two mid-to-senior full-stack engineers. The old process: a degree filter, a recruiter screen on years of experience, then a four-round loop. They got 400 applicants, interviewed 12 who looked right on paper, and hired one, who struggled with their TypeScript codebase.
The skills-based version: the job post lists three outcomes and four skills. Every applicant who passes a two-minute knockout (work authorization, time zone) gets a 40-minute work sample: fix a failing endpoint in a small TypeScript repo, AI tools allowed, then write five sentences on what they changed and why. Two engineers score it against anchors in 10 minutes each.
Of 400 applicants, 160 take the test, 22 clear the bar, 8 go to interviews, and they make two offers. Three of the 22 had no degree; one of those turns out to be the strongest engineer of the round. The cost was about 50 hours of engineer time spent grading, which is real money. They decided it beat interviewing 12 people chosen by logo.
The Honest Counterpoint
Skills tests have their own bias. A 40-minute unpaid exercise is easy for someone between jobs and hard for a parent working full time. Keep tests short, pay for anything longer than an hour, and offer flexible deadlines.
Tests also miss things. They measure what fits in a box: code, written reasoning. They don't measure judgment on a six-month project or how someone behaves when production is down at 2am. References and a trial period still matter. And for genuinely specialized roles, say cryptography or a compiler team, deep domain background may be the best evidence you have, and treating it as "pedigree bias" would be a mistake. Read hiring for potential over pedigree for where that line sits.
Frequently Asked Questions
What is skills-based hiring for software engineers?
Hiring decisions based on demonstrated ability, from work samples, coding tests and structured interviews, rather than degrees, employer names or years of experience.
Does skills-based hiring actually work?
It works when the process changes along with the job post. The Harvard and Burning Glass research found that firms that really followed through saw non-degreed hires stay longer: their retention rate was 10 percentage points higher than that of degree holders.
How do I start skills-based hiring with a small team?
Pick one open role. Write four skills, build one short work sample, score it with anchors, and look at resumes only after the test. Or start from a bench where the test already happened, like pre-vetted developers.
The Bottom Line
Skills-based hiring is a process change, not a policy statement. Test early, score with anchors, allow the tools engineers really use, and look at pedigree last.
See blind profiles of engineers who passed, or describe your role to Sol.
Roberto Espinoza is CEO of Ruzora, which helps US startups hire pre-vetted senior LATAM engineers, with a vetted shortlist in 72 hours. See available engineers.
