I went through a five-stage interview process with DataSnipper, including a take-home technical assessment, with positive feedback from the recruiter after each round. Based on the experience, I’d encourage future candidates to think carefully about how much time they invest.
The process runs roughly like this: an initial recruiter screen, a hiring manager interview, then a take-home technical assessment. After that comes an on-site assessment round where you present and defend your work, followed by separate interviews with the Director of Engineering and the VP. The recruiter stays in touch throughout and relays feedback between stages.
The technical assessment was my main concern. The company provides no resources for it, so you rely on your own paid subscriptions (Claude, Figma for programmatic access, etc.) to build a working POC. Beyond the core business problem and prototype, they also expect a presentation on framework implementation and adoption strategy. A significant amount of (unpaid) professional time. I made a real effort to keep the setup simple to run locally (via Docker and dedicated local artifactory), but it became clear the work wasn’t closely reviewed: I got no follow-up questions, and as far as I could tell only one person had opened the submission. I would have welcomed feedback of any kind, including critical, on the approach or how I’d extend it. None came.
The interviews themselves felt inconclusive. As someone who also interviews candidates for AI roles, I have a sense of when a process is well structured, and this one didn’t feel that way. Rather than digging into the assessment I’d built, or working through case scenarios, system design, or business-oriented problems, the conversations went elsewhere.
The VP, for instance, asked how we run agentic SDLC at my current company, unexpected for the position, and as someone bound by NDAs I could only answer at a vert high level, which made for an awkward exchange (at least on my side). I’d have expected questions tied to the role itself: the assessment problem, or how to build a design framework and drive adoption across teams.
There was also a noticeable disconnect within leadership. Across all five stages, no two interviewers seemed to share the same understanding of the role, and few aligned with the original job description. The one interviewer who articulated a clear, value-driven business problem was non-technical. The technical leadership focused on ambitious end-to-end systems. Automatically fixing bugs, building applications from scratch, that would realistically need a dedicated team to validate, while the foundational data structures, development processes, and AI governance to support them didn’t yet appear to be in place.
The technical leads also seemed unaware of issues the hiring manager had flagged earlier, such as inconsistency across engineering teams and low adoption of their design language (something visible just from auditing their public website across pages). I’d have expected a discussion of how those concrete problems might be addressed, or a system design conversation with the Director of Engineering, including a hypothetical framework and rollout process to bridge the teams. None of that materialized.
Ultimately, I was told the feedback was highly positive but that the role's seniority requirements had shifted at the final stage. I was offered an undefined alternative role that would require restarting the entire loop from scratch. Given that the original assessment was already full-stack, a pivot to a more targeted backend/AI discussion would have been a reasonable compromise if there was genuine alignment.
My takeaway: roles shouldn’t be opened before there’s internal consensus on what they are. My advice to candidates, protect your time, and don’t over-invest in the assessment.
