How We Revamped Hiring at CTA: Human Values First, AI as an Amplifier
By: David Spearman
At Community Tech Alliance (CTA), we believe technology should always serve people, not the other way around. But that principle doesn’t just apply to the software we build or the data infrastructure we develop; it applies to how we run our own organization.
Over the past few years, CTA has made a multi-year investment in our Diversity, Equity, Inclusion, and Accessibility (DEIA) practices through what we call our Cultural Propagation Practice. We recognize that culture requires intentional, ongoing nurturing. Our practice empowers everyone at CTA to be active stewards of our shared values in perpetuity.
During our last company retreat, we turned a critical lens inward and asked a fundamental question: What are the most crucial places we can propagate our culture?
One answer was clear: How we show up before someone even works here.
Our hiring and onboarding processes are the very first touchpoints a prospective employee has with CTA. To ensure those touchpoints genuinely reflect our commitments to belonging, transparency, and equity, we formed a working group of four team members to audit, reimagine, and revamp how we hire.
We recently put this redesigned process to work for a DevOps Engineer role, as we test out small, iterative changes to larger processes. Here is what we changed, how we used AI thoughtfully behind the scenes, and what we refused to outsource to technology.
1. Redesigning the Experience to Reflect Real Work
In many traditional hiring processes, interview rounds are created in silos. One interviewer writes their own questions, another asks overlapping details, and candidates are evaluated on an ambiguous "vibe check." We wanted a process that respected candidate time and truly reflected what day-to-day life at CTA feels like.
Here are four key changes we made to our interview process:
A Holistic Script Audit: We reviewed every interview step and script across the entire process to ensure we were evaluating critical skills without asking the candidate the same questions multiple times in different rounds.
Simulating the Real Working Environment: Rather than asking only boilerplate interview questions, we invited candidates into transparent conversations about the actual technical and personnel challenges our team is navigating. During technical panels, candidates engaged with the team in a collaborative working session, giving and receiving constructive feedback on their technical exercise much like they would during a design doc review or peer code review.
Criteria-Based Evaluation Forms: We replaced subjective overall "yes/no" impressions with structured rubric forms in our Applicant Tracking System (ATS). Every interviewer now evaluates candidates against explicit, objective criteria.
Adapting to Team Resources and Capacity: Equity also means respecting our internal team’s capacity. For our DevOps role, we shifted the Hiring Manager interview earlier in the sequence than for other roles. Our engineering team was focused on critical delivery deadlines, and this adjustment ensured we only asked the broader team to sit on a panel interview for our top finalists, protecting team capacity while maintaining a thorough process for applicants.
2. Using AI to Enhance (Not Replace) Human Engagement
As a tech organization, we aren't shying away from AI tools like Claude and Gemini, but we approach them with clear boundaries. In our hiring revamp, AI acted as an administrative amplifier: helping us check our work, maintain consistency, and reduce bias.
How We Used AI in the Process:
Auditing Old Scripts: We fed our historical DevOps interview scripts, job description, and CTA values documentation into Gemini to flag gaps where we were under-assessing key skills or unintentionally repeating questions across rounds.
Refreshing Content with Nuance: Gemini helped draft updated interview questions tailored to the specific technical nuances this role requires today.
Building Structured Evaluation Rubrics: Once our scripts were set, we used AI to draft objective rubric criteria to populate our ATS evaluation forms.
Synthesizing Feedback Trends: We leveraged our ATS’s built-in AI insights feature to summarize trends across multiple interviewer evaluations, helping us highlight clear candidate strengths and surface follow-up areas for future rounds.
Crucially, every prompt, script, and rubric generated by AI started with human intent and ended with thorough human review and editing.
3. Our Non-Negotiables: Where AI Has No Place in Hiring
While AI is great at spotting patterns or helping draft rubrics, it lacks human empathy, context, and judgment. We established strict guardrails around what technology is never allowed to do in CTA's recruitment:
No AI Resume Screening: We do not use automated AI resume screeners or algorithms to weed out candidates. Every single application submitted to CTA is reviewed by a real human being. (While we use basic objective screening questions, auto-disqualifications are never made by an AI judgment call.)
No Interview Recording Bots: We do not use AI bots to record, transcribe, analyze, or evaluate candidate body language or speech during interviews. All notes and evaluations are generated by our employees.
No AI Candidate Communications: Every email, phone call, and video interview engagement with a candidate comes from a human team member. We never allow an AI agent or bot to communicate directly with candidates.
Culture as an Ongoing Practice
By grounding our hiring overhaul in our Cultural Propagation Practice, we’ve created a hiring experience that is fairer for applicants, more efficient for our internal team, and deeply aligned with CTA’s core values.
Using AI thoughtfully allowed us to double-check our work and maintain consistency, but human care remains the foundation of how we build our team.
CTA is constantly growing and evolving! We are currently hiring for a Senior/Staff Software Engineer role. Check out our Careers Page to learn more about our open positions and experience our hiring process firsthand.