Tuesday, 19 February 2019

Effective Technical Interviewing with GitHub

Throughout my career I've encountered many problems with hiring processes, some mildly annoying, some decidedly infuriating, and some utterly baffling. Of all of those issues, few have pressed my buttons as consistently as those relating to how technical assessments are generally handled.

The cost of an ineffective technical interview to the company can be greater than the hours interviewing or the lacking resources until the right candidate is found, with many regulations making it difficult to get rid of a low-performing employee and most situations requiring a learning curve that extends beyond legal probation periods.

The cost of a technical interview to the applicant can be even greater, as interviewing at multiple companies or when already employed leaves little time to do the assessments and it's usually a lot of effort, for no compensation, that produces nothing of value.

The bigger companies have found a reasonably good solution to the technical assessment, but phone screening and days full of whiteboard interviews are simply not affordable for the smaller players. The focus of this article is on technical assessments wherein the candidate is expected to produce a solution, or set of solutions, within a set period of time.

There are two major downsides to these coding assessments:

1. It is of little consequence whether this is done in an in-office setting or as a take-home assignment, it is very difficult to extract meaningful information about the strengths and weaknesses of a candidate from the results of an assessment done in this manner. There are many variables that may influence the results of this kind of an assessment, the candidate's ability to code under stressful conditions for a start, and there's not usually enough interaction with the candidate during the coding process to assess collaborative behaviour.

2. The work done in producing these solutions is, for all intents and purposes, wasted effort. Candidates are consistently asked to demonstrate their skills by producing arbitrary code that would not be of use or interest to anyone going forward.

In this article I'd like to address two possible strategies using GitHub that could significantly improve the effectiveness and overall outcomes of this important step in a candidate's application.

First strategy: Existing GitHub issues

The world of open-source is full of real problems that require solving, and nine times out of ten (for smaller organizations in particular) in software packages that are used by the interviewing company.

One possible strategy is to break down the technical skills you require from your candidates - not just the languages and frameworks, but problem spaces as well - and find open issues on GitHub with similar profiles.

Second Strategy: Porting an issue to GitHub

It is a common misconception that everything in a proprietary codebase must be kept under lock and key. With the right license, there are plenty of issues that a company might have that can be isolated from the codebase, ported to GitHub, and then reincorporated once solved.

Similarly to the first strategy, it is important to break down the technical skills you require from your candidates and find issues that would highlight their relevant strengths and weaknesses. The act of "open-sourcing" this code would be as simple as establishing the right license and creating a repo. Ideally, it would be code that others might find useful, too!

With this strategy, there would need to be a different problem for each individual candidate, as the moment one candidate has come up with a solution there would be little to stop another from taking a peek and little value in multiple developers repeating the same effort.

The advantage, obviously, is that work put into interviewing would directly benefit the company, but the moment you're benefitting from the candidate's efforts they would be entitled to at least some compensation, in addition to being able to show their work in other job applications.

When an applicant has successfully delivered a solution, you could compensate them directly, which may be preferable, or you could use Gitcoin, where you can set a bounty on your issues for immediate payment on the acceptance of a PR.

Either way, it'd be a small price to pay for development, the candidate would have earned a little money (which most candidates desperately need), and you'd have useful information for your hiring process.

Conclusion

Contributing to GitHub projects will show more than just a candidate's code quality; it can also show communication skills and collaborative behaviour. Additionally, choosing or defining projects with strict requirements for code conventions and unit tests will push your candidate more significantly than just mentioning that you'd like to see testing done.

An accepted PR (pull request) will make the world a (slightly) better place; it would be a good deed on your part, a good deed on the candidate's part, and the resulting effort would have actual value.

It is also far more likely that a candidate will be motivated to solve a real problem than a fictitious one, and their efforts will be rewarded by becoming a part of their portfolio. As somebody who has predominantly worked on proprietary code, I can testify to the utility of being able to direct a potential employer to my GitHub account!

With either strategy, everybody wins. For the developer, the technical assessment becomes real world experience that they can show off to others, and the employer gets to see real-world coding performed under real-world conditions.

Wins all around!

No comments:

Post a Comment

@mysql/xdevapi joy!

after hitting a wall last night with the existing node.js mysql packages, which for some reason have been years behind integrating with mysq...