My candidate experience was disappointing, and the biggest issue was not difficulty — it was opacity. The public job description described a role centered on enterprise AI solutioning, technical discovery, executive communication, evaluation frameworks, and practical deployment strategy. That is exactly the kind of work I expected to be evaluated on. Instead, the process felt under-calibrated, overly opaque, and too dependent on candidates reverse-engineering unstated expectations. The early stages did not reflect the true candidate burden. The real investment came later, where preparation became a multi-hour effort across multiple days. I do not object to hard assessments. For an Applied AI role at a frontier AI company, the bar should be high. But when a company asks candidates to invest that much time, the evaluation criteria should be explicit. The later-stage assessment appeared to test many dimensions at once: solution design, technical communication, executive-level framing, business impact, implementation planning, product fluency, ambiguity management, and live objection handling. Those are all relevant to the role. The problem is that I was not given a clear rubric for how those dimensions would be weighed. Tactical prep is not the same as real calibration. That created what felt like a hidden-rubric problem. Candidates are asked to be creative, technical, concise, consultative, enterprise-aware, product-aware, and executive-ready — but without enough clarity on what “great” actually means. In my view, that rewards candidates who are best at guessing the interview format, not necessarily candidates who would perform best in the role. The process also seemed to blur the line between assessing Applied AI judgment and expecting internal product fluency from an external candidate. There is a real difference between testing whether someone can reason through technical customer problems and testing whether they already know product-specific answers that employees would normally learn through internal enablement. External candidates should not be evaluated as though they have already been onboarded. The recruiting experience was professional but too transactional. I received polite communication and some useful tactical guidance, but not the level of expectation-setting or transparency I would expect for a process requiring this much candidate investment. Candidates should ask direct questions early about evaluation criteria, decision-making, leveling, compensation philosophy, and what flexibility exists later. Do not assume clarity will come at the end. The lack of feedback was also frustrating. The company acknowledged that the process required real work, but after the interview there was no substantive feedback. I understand limited-feedback policies for standard interviews. But for a custom, high-effort assessment, providing no meaningful feedback makes the process feel one-sided: candidates are expected to give a lot, while receiving very little transparency in return. My advice to future Applied AI candidates: prepare deeply around enterprise AI deployment, architecture, data handling, hosting, security, governance, customer implementation, rollout strategy, stakeholder alignment, and measurable business impact. Do not prepare like this is a generic technical interview or a standard pitch. Prepare to show how an AI solution would actually work in a real enterprise environment — and be ready for highly specific technical objections. My takeaway: take the opportunity seriously, but protect your own time. Brand prestige does not guarantee a transparent, well-calibrated, or candidate-centered hiring process. In my experience, this process placed too much burden on candidates to infer expectations that should have been made explicit.