| Hey HN, The way we hire engineers made sense in 2015. But in 2025, when engineers use AI tools daily, we're still testing algorithm memorization on whiteboards.. That's why we're building DevDay. DevDay is built for the new reality of modern engineering work: candidates collaborate with AI teammates, delegate tasks to AI agents, and solve problems using the tools they'd actually use on the job (LLMs, git, and Slac(k) for team communication). The old interview playbook is fundamentally broken:
- Whiteboard anxiety tests don't predict performance
- Take-home tests and virtual paired programming get gamed with ChatGPT
- Algorithm memorization has zero correlation with debugging prod issues (what you actually deal with in your day to day work) Here is what we are not:
X Another LeetCode clone with AI buzzwords
X Replacing engineers with AI
X "Disrupting" hiring with magic algorithms What it actually does:
- Tests AI collaboration skills (AI teammates, delegate task to agents, coding assistant integrations)
- Simulates real team environments and workflows
- Shows problem-solving approach, collaboration and behavioral skills, not memorized solutions
- Assesses how candidates think and communicate Questions for HN because we are genuinely curious: - Do you assess engineers who work with AI daily? If yes, how do you do it today?
- What would technical interviews look like if designed today within your organisation?
- Are we testing skills that matter in 2025? Link: trydevday.com P.S. - Yes, someone will suggest "just pair program" or "check their GitHub." Great for small teams, doesn't scale when hiring 10+ engineers monthly. |
This got strong "rise and grind guys thinking about life hack" vibes pretty quickly. Be serious, who's hiring at that rate currently?!
I see whiteboard interviews as an answer to disputes over opinionated tooling. It's just a pen, a blank canvas, and natural language to communicate. There's probably just pseudocode, no IDEs and zero runtimes.
Bringing "AI teammates" into the mix reintroduces some disputes: Candidates would lack the experience to get around your tool and trigger the right responses. Different LLMs have different "characters". As an engineer you'd want to pick the best tool for the job, and no engineer would like be stuck with your choice. It's usually an effort to figure out and setup such a tool for each distinct project.
Besides that, even technical interviews have the social component to rule out any "cultural" differences within the team. I really doubt that good technical leaders could take that much value out of an AI assessment to just skip any in-person step of the interview.