How a Tech Company Built a Diverse Engineering Team

The Challenge

A 300-person tech company had a diversity problem they knew they needed to solve. Their 80-person engineering department was 82% male and drew 67% of hires from the same 5 universities. The CTO recognized that this homogeneity was not just an equity issue - it was limiting the team's ability to build products for a diverse customer base and reducing the range of problem-solving approaches in technical discussions.

Previous diversity initiatives had produced limited results:

• Targeted job postings: Posting on diversity-focused job boards increased application volume from underrepresented candidates by 15%, but the same resume screening process filtered most of them out. The screeners - unconsciously - favored candidates from familiar universities and companies.
• Blind resume review: Removing names and photos helped with initial screening but did not address the deeper issue: the evaluation criteria themselves favored traditional tech backgrounds (CS degree, FAANG experience) over equivalent but non-traditional paths (bootcamp graduates, career changers, self-taught developers).
• Diversity recruiters: An external diversity recruiting firm produced candidates, but at 25% placement fees, the cost was unsustainable for ongoing hiring. And the hires were concentrated in junior roles because the firm's network skewed early-career.

The Solution

WorkSwipe's bias-aware matching engine was fundamentally different from both traditional screening and simple blind reviews. Rather than hiding demographic signals and hoping for better outcomes, the system actively debiased the evaluation criteria themselves.

Skills-based matching replaced credential-based filtering. Instead of using university name, company pedigree, and years of experience as primary filters, WorkSwipe evaluated what candidates could actually do. The system analyzed code contributions, project complexity, problem-solving patterns, and technical depth independent of where or how candidates acquired those skills. A self-taught developer who built production systems was ranked alongside - and often above - a CS graduate who had only worked on internal tools.

Expanded source reach. Traditional recruiting targets the same talent pools that every other company targets. WorkSwipe expanded sourcing to include coding communities, open source contributors, bootcamp alumni networks, and professional communities for underrepresented groups in tech. The AI identified high-potential candidates in these pools who would never have appeared in a standard LinkedIn search.

Structured evaluation reduced subjective bias. Every candidate was evaluated against the same criteria using the same rubric. Interview questions were standardized, and interviewer feedback was collected through structured forms that asked about specific competencies rather than "culture fit" (a term the company retired because it had become code for "similar to us").

The Results

Over 12 months, the engineering team's composition shifted meaningfully:

• Representation: Underrepresented groups in engineering grew from 18% to 41% of the department
• Source diversity: Hires came from 23 different universities and 7 different entry paths (traditional CS, bootcamp, self-taught, career change, military transition, community college transfer, and international relocation)
• Quality maintained: Average technical interview scores remained statistically identical (4.1 vs 4.0 on a 5-point scale), confirming that broader sourcing did not require lowering the bar
• Retention: 94% of diverse hires were still with the company at the 12-month mark, compared to 88% for the overall engineering team - disproving the common concern that diversity hires are less likely to stay
• Innovation impact: The team's internal innovation score (measured by cross-team collaboration, novel solution adoption, and patent applications) increased 22% year-over-year
• Candidate experience: NPS score from candidates who went through the process improved from +12 to +47, with candidates specifically citing the fair and transparent evaluation process

Why It Worked

Changed the criteria, not just the pool. Most diversity hiring efforts focus on finding more diverse candidates and putting them through the same biased process. WorkSwipe changed the process itself - evaluating skills and potential rather than credentials and pedigree. This made the entire hiring pipeline more equitable by default.

Data replaced intuition. "Culture fit" interviews are where unconscious bias does the most damage. Structured evaluations with specific criteria and scoring rubrics reduced the space for subjective judgment while actually improving hire quality. The data showed that structured processes selected better candidates regardless of background.

Inclusion supported retention. Hiring diverse candidates is only half the challenge - keeping them requires an inclusive environment. The company paired the hiring changes with mentorship programs, ERGs (Employee Resource Groups), and inclusive meeting practices. The 94% retention rate reflects both good hiring and good support.

Key Takeaways

• Debias the criteria, not just the data. Hiding names on resumes is necessary but insufficient. The evaluation criteria themselves need to measure skills and potential rather than proxies like university prestige and employer brand.
• Broader sourcing does not mean lower standards. When you evaluate on actual skills, candidates from non-traditional backgrounds perform as well as those from traditional paths. The pipeline was the bottleneck, not the talent.
• Measure everything. Track representation, quality, retention, and candidate experience separately. Improvements in one metric at the expense of others is not sustainable progress.
• Hiring changes need culture changes. AI can fix the pipeline, but retention requires human investment in mentorship, belonging, and growth opportunities for all team members.