Certification program

DevScanr analyzes public profiles on GitHub and identifies the best developers. Our mission is to change the industry, so people are hired for what they do, rather than for what they say.

Our certification program helps developers to present their open-source work and achievements in a form that is readable to tech recruiters. Namely, in the form of an extended individual certificate.

Motivation

The data on LinkedIn or in a personal resume is not always reliable. Jobseekers exaggerate their experiences and tweak titles to apply for better positions. Recruiters are aware of that and increasingly require proofs of competence. Endorsements, testimonials and certificates serve this purpose, to different extents.

Certificates are the strongest evidence of qualification, because they are issued by a neutral, unbiased party. According to the U.S. Bureau of Labor Statistics, having certificates and licenses in your resume can increase median earnings by up to 30%. Or, with a fixed salary, one can receive a job offer much faster.

Many well-known certification programs are expensive and require significant effort to prepare for. DevScanr rewards for the work that is already done, therefore our certificates are free and take no extra time.

Open-source contributions and activities deserve recognition and respect. Unfortunately, it is largely ignored by recruiters who do not have the necessary skills to read and evaluate codebases. We aim to change that by consolidating industry-specific data into a readable and understandable form.

How it works

The qualification process starts with an automatic analysis of public GitHub repositories, contributions, activities, social metrics, ratings, etc. by dedicated ML algorithms. The goal is to extract valuable insights of dev. capabilities and preferences. Unlike resume, such open data is very difficult to forge due to omnipresent timestamps. Each certificate is finally approved by a human subject-matter expert from our team.

Note: nowhere in the project we rely on Chat-GPT or other LLM (large language models) solutions. We build and support a proprietary ML-based expert system. It is multi-faceted and relies on neural networks for certain tasks. Yet, the whole thing is specialized, does not hallucinate, and is much more performant than anything LLM-based.

So we analyze repositories... By no means the amount of open-source activity is equated to performance. Most programmers have the majority of their work in private, and we totally recognize that. Above average open-source activity does not necessarily mean an above average engineering potential. Still, to be fair, any objective proof of capabilities should be preferred over barebone text-backed claims.

One more foundation of this project is crowd-assessment. Earned stars and followers are considered in our conclusions and affect them. While we do not rank developers by mere volume of published or earned artifacts, social proofs are invaluable. Community evaluation is not perfect, but it is, arguably, more objective and trustworthy than any alternative. There’re more angles to this. DevScanr utilizes graph algorithms (think Google PageRank) to extract knowledge from GitHub “follows” relations. We post some of our findings and decisions in our Blog.

Limitations

As noted, our analysis is exclusively based on public repositories, contributions, and interactions. So it’s one natural limitation. We are also limited by the quality of data and algorithms. There’re many of them and a lot of math is involved. Not all skills can be properly recognized, no competency can be perfectly quantified – we acknowledge that and apologize for all potential mistakes and imperfections. DevScanr team continuously improves the quality of the product, and our results will be more precise and refined over time.