How an AI recruiter underestimated a Rust developer because of an old profile

A Rust developer submitted a resume for a Senior position. The screening system used by the recruiter read the digital footprint and delivered a verdict: Junior+/Middle. The problem turned out to be that the AI encountered an old profile on the internet — a profile from when the experience was actually 1.5 years. Over several years, much happened: two large projects reached release, skills grew, but the first trace remained in search indexes.

Why AI Recruiters Fail

Modern candidate screening systems rely on open data: profiles on freelance platforms, GitHub, LinkedIn, old resumes on recruitment sites. The problem is that these sources update slowly or not at all. The AI takes everything it finds and tries to compose a portrait of the developer. When multiple versions of a profile are found — old, medium, new — the system can weigh all sources, but often the old information has greater impact. Ratings that freelance platforms assigned three years ago still live in search results and participate in calculations. The algorithm perceives the age of indexing as a sign of reliability: the longer information has been on the internet, the more stable it is.

The Problem of Credentials

• Code on GitHub can be private — the AI cannot see it
• LinkedIn updates slower than real professional growth
• Old resumes on Habr and other boards remain in search forever
• The AI cannot assess real contribution to closed company projects
• The number of stars on GitHub does not correlate with salary level or seniority

There is no fault in the AI itself — this is an architectural problem. The system works within the bounds of available data. When the digital trace is contradictory and outdated, errors are inevitable.

How Recruiters Solve This

Agencies and large companies use two-level screening: AI filters out obvious non-fits, then a live recruiter manually checks questionable profiles. On freelance platforms and in startups, this luxury often does not exist. There the system gives a final verdict, and the candidate either passes or does not. The Rust developer could dispute the decision with a live person, but the recruitment platform might not provide such a channel. The alternative is to update open profiles yourself, add new data, create a GitHub project that demonstrates your level. All this requires time.

What This Means

AI systems exacerbate the problem of information inertia. If you have grown professionally but old information is still on the internet, the system will hinder you from moving forward. This is not a reason to abandon AI screening — rather, it is a reason to be more careful about your digital footprint and regularly update the profiles visible to the world.