Refresh your technical recruitment methods

You are looking for that engineer that seems to be impossible to find. As a tech recruiter you just sometimes get stuck in finding that right candidate. 

You know the tricks that everybody knows but you want to refresh your sourcing strategy and methods.

The cross platform playbook still works, but the platforms underneath it have moved. GitHub passed 180 million developers and now adds roughly one new account per second - GitHub Octoverse 2025. Stack Overflow went the other way: only 3,862 questions were posted there in December 2025, a 78% drop year on year, against a peak of over 200,000 a month in early 2014 - DevClass.

So the ideas below are the ones that still hold, with the parts that have changed marked as such.

1. Source from the unconventional places

As you probably know there are many other places than LinkedIn to find candidates on.

With a cross platform sourcing strategy you systematically search for candidates across several professional and social media platforms.

This involves a lot more platforms than LinkedIn. Think about platforms like GitHub, Wellfound (the startup hiring platform that was spun out of AngelList and renamed in November 2022, now claiming 10M+ startup-ready candidates and 25k+ companies), Stack Overflow and Kaggle. Just to name a few.

Technical recruiters that source on alternative platforms have big benefits like less competition, richer candidate information and access to niche players.

One correction worth making, because it changes how you use it: Stack Overflow is now an archive, not a feed. With question volume back at 2008 levels, "answered three questions last month" is no longer a signal you will find. What you will still find is fifteen years of answers, and an answer written in 2019 that explains Kubernetes networking clearly is still evidence that the person understood Kubernetes networking. Search it for depth, not for who is active right now. For that, GitHub is where the activity went.

These are the best platforms to find tech talent on.

2. Fix the contact problem before you scale the search

Here is why most recruiters try GitHub sourcing for two weeks and then quietly go back to LinkedIn: finding the engineer is the easy half. You end up with a username, a profile picture of a cat, and no way to say hello.

The classic trick still works sometimes. Public commits carry the author's email, so you can open a developer's personal repo, add .patch to a commit URL, and read the address they committed with. The catch is that GitHub rewrites that address to ID+username@users.noreply.github.com whenever the author has kept their email private, and that is the format for every account created after 18 July 2017 - GitHub Docs. On a developer who joined in the Copilot era, expect the noreply. On someone with a ten year old side project, try the oldest repo first.

When the trick fails, you are into contact databases, and it pays to know what you are actually buying. The two most recruiters end up on are Apollo.io and Lusha. Both sell the same thing: a lookup from a person plus a company to an email address and often a phone number. Both were built for salespeople, which is the detail that decides whether they work for you.

Lusha does the same job from a Chrome extension at $49.90 per user per month for its entry tier, and is worth a look if your sourcing happens mostly on profile pages rather than in bulk lists. Whichever you pick, verify before you send: a bounce rate above a few percent damages the sending domain, which quietly buries the emails that were deliverable. The mechanics of that are in our guide on finding candidate email addresses.

3. Search on the signal, not on the job title

The reason alternative platforms give you richer candidate information is that they index behaviour instead of self description. A LinkedIn headline is a claim. A repository is evidence. That difference is only useful if you search on it.

GitHub's user search takes qualifiers you can combine, all documented publicly - GitHub Docs. A query like type:user location:berlin language:go followers:>50 gives you Berlin based accounts whose repos are mostly Go, with enough of a following to suggest other developers rate their work. Add created:<2016-01-01 when the role genuinely needs someone senior, and type:user always, or half your results will be company accounts.

Two honest limits on this. Location is a free text field, so it is full of "remote", "127.0.0.1" and country flags, and filtering on it will silently drop good people. And followers rewards fame, not skill: a developer who wrote one viral tutorial outranks the quiet person maintaining a library your stack depends on. Use the qualifiers to build a long list fast, then judge the long list by reading the actual code and the actual issues someone files, which is the part no filter does for you.

4. Re-source your own database before you source the internet

The most underused sourcing channel in technical recruitment is the ATS you are already paying for. Every closed engineering role leaves behind candidates who reached the final round and lost by a hair, and most of them are never contacted again.

They are a better starting point than a cold GitHub list for three reasons. They already understood what your company does, so you skip the pitch. They already passed a technical screen your own team designed, so you are not guessing about the bar. And the reason they lost is often gone: they were three years short on experience, or you had one headcount and two finalists. Twelve months later, neither is true.

Make it a habit rather than a campaign. When a req opens, search the ATS for the last two years of applicants to similar titles before you open a single sourcing tab, and message the finalists first, naming the role they nearly got. That is a warm message you can send in five minutes and it costs nothing.

5. Let an agent take the first pass

The genuinely new idea since this article was first written is that the long list step can be handed off. AI sourcing agents now do the search, the cross platform matching and the first outreach message on their own, which changes what a recruiter's day looks like: you spend it on the ten people worth a conversation instead of the four hundred worth a filter.

Tools in this class include HeroHunt.ai, which searches across platforms rather than a single database and handles the outreach automatically, alongside other AI sourcing tools you can compare in our overview of AI sourcing tools. Treat any of them as a first pass and not a verdict. An agent is very good at finding forty plausible Go engineers in Berlin and still cannot tell you which of them will survive your architecture interview.

Where to start

If you take one thing from this: the bottleneck in technical sourcing is almost never the search. It is the contact details and the follow through.

So the order that works is boring on purpose. Re-source your own ATS first, because it is free and warm. Then run one GitHub query with real qualifiers instead of ten LinkedIn searches. Then solve the address problem properly, with a tool and a verification step, before you scale anything. Use Stack Overflow for evidence of depth rather than signs of life, and give an AI sourcing agent the long list work you were never going to enjoy anyway.

Written by Yuma Heymans (@yumahey), who built HeroHunt.ai, the world's first AI Recruiter, now used by 15,000+ recruiters to source and contact candidates across platforms on autopilot.