The hiring manager was absolutely gutted. After three rounds of interviews, their top candidate, a brilliant data scientist whose technical assessment had blown everyone away, had withdrawn from the process. When I asked why, the answer was both simple and maddening: "He asked if he could take the behavioural interview questions in writing rather than face-to-face. HR said no because it wouldn't be fair to other candidates."
And just like that, they lost someone who could have transformed their ML pipeline, all because of a rigid interview process that couldn't accommodate a candidate's neurodivergent traits.
I've watched this happen repeatedly across my eight years in ML engineering and recruitment. Companies claim to value different thinking but design hiring processes that systematically filter it out.
The Neurodiverse Talent Pool You're Missing
Let me be specific about what we're talking about here. Neurodiversity encompasses conditions like autism, ADHD, dyslexia, dyscalculia, and dyspraxia. These aren't rare edge cases, they're remarkably common in tech, and particularly in ML and data roles.
But the standard tech hiring process is practically designed to exclude these candidates. Open-plan assessment centres, ambiguous questions, timed tests with vague instructions, unstructured interviews where "cultural fit" becomes code for "thinks exactly like us." I've sat in dozens of interview panels where the phrase "bit odd" was enough to sink a technically brilliant candidate.
This isn't just morally questionable. It's commercially stupid.
Some of the most innovative thinkers I've worked with, people who could spot patterns in data that others completely missed, who built ingenious model architectures, who questioned fundamental assumptions, were neurodivergent. Their different cognitive wiring was precisely what made them exceptional at complex technical tasks.
Beyond Tick Boxes: Diversity Recruitment Strategies That Actually Transform UK Workplaces
Master the Virtual Hot Seat: 7 Video Interview Techniques Recruiters Don't Tell You
How to Master 'Tell Me About Yourself' Interview Question: UK Expert Insights
What Neurodiverse Talent Actually Wants (Hint: Not Special Treatment)
I spoke with a former colleague last week, one of the most talented ML researchers I know. He's autistic and has ADHD. "I don't want special treatment," he told me. "I just want the chance to demonstrate my skills without having to first prove I can handle fluorescent lights and make small talk with strangers."
This gets to the heart of the issue. Most neurodivergent candidates aren't asking for lower standards. They're asking for different pathways to demonstrate the same skills.
The accommodation requests I've heard most frequently:
- Clear, specific instructions (rather than vague, open-ended tasks)
- Written communication options for verbal tasks
- Quiet assessment environments
- Direct, explicit feedback
- Flexible timing for technical assessments
- Advance sharing of interview questions
- Permission to use notes or reference materials
None of these compromise the quality of your hiring. They simply allow candidates to showcase their abilities without irrelevant barriers.
The Updated Job Ad: Signalling From Day One
Your job advert is the first barrier to entry. Take a look at yours now. Does it include language like "excellent communication skills," "team player," "thrives in fast-paced environments," "must handle ambiguity well"?
These phrases are red flags for many neurodivergent candidates, not because they can't work in teams or communicate, but because this language often signals an environment that won't accommodate different cognitive styles.
Some practical changes:
- Be specific about what "communication skills" actually means in the role
- Separate essential from nice-to-have requirements
- Include an explicit statement about accommodations
- Focus on outputs (what the person needs to accomplish) rather than style (how they work)
For example, instead of "excellent verbal communication skills," try "able to explain technical concepts to non-technical stakeholders through documentation and presentations."
Technical Assessment Redesign: Let Skills Speak
The technical assessment is where you can level the playing field most effectively. Traditional approaches, like whiteboarding algorithms while being watched, or timed Hackerrank challenges without clear instructions, create artificial barriers.
A client of mine recently redesigned their ML engineering assessment. Instead of a high-pressure, 90-minute live coding session, they now offer three options:
- A take-home project with clear requirements and a generous timeframe
- A portfolio review with discussion of past work
- The traditional live assessment (still available for those who prefer it)
The results were striking. Not only did they see more neurodivergent candidates progress, but the overall quality of hires improved. Why? Because people could showcase their abilities in the format that worked best for them.
The key is optionality. Different brains work differently. Some people genuinely perform better under time pressure. Others need quiet processing time. By providing options, you let candidates put their best foot forward.
The Interview: Where Most Processes Break Down
Interviews are social performances. That's the problem. Traditional interviews often assess social conformity rather than job capabilities, and that's where many neurodiverse candidates get filtered out.
Last month, I advised a London fintech on overhauling their ML team interview process. Here's what worked:
- Providing interview questions 24 hours in advance
- Offering a choice between video or in-person formats
- Allowing candidates to submit written responses to some questions
- Training interviewers on neurodiversity
- Focusing on specific examples rather than hypotheticals
- Explicitly stating that eye contact isn't expected or evaluated
These changes didn't lower standards, they just stopped penalising candidates for differences that had nothing to do with job performance.
The Legal Bit: Avoiding Discriminatory Practices
It's worth noting that the UK's Equality Act has been increasingly interpreted to provide protection for many neurodevelopmental conditions. The most recent case law (as of early 2026) has strengthened these protections.
But the more compelling reason to adapt your process isn't legal compliance, it's competitive advantage. In the scrambling fight for ML talent that I've been witnessing over the past 18 months, companies can't afford to arbitrarily exclude qualified candidates.
What Works: Practical Steps from Companies Getting It Right
I've had the privilege of helping several UK tech companies redesign their hiring processes to be more inclusive of neurodivergent candidates. The ones that succeed share a few common practices:
- They offer flexibility in their assessment methods
- They provide clear, explicit instructions
- They focus on the outcomes, not the process
- They train their interviewers on neurodiversity
- They ask candidates what accommodations would help them perform at their best
This last point is crucial. Sometimes we overthink this stuff. Just ask candidates what would help them showcase their abilities.
Beyond Hiring: Creating a Genuinely Inclusive Workplace
There's not much point in adapting your hiring process if your workplace itself remains hostile to neurodivergent colleagues. I've seen too many companies make this mistake, they hire neurodivergent talent only to lose them within six months.
The most common post-hiring adjustments that actually retain neurodivergent talent:
- Flexible work arrangements (remote options, quiet spaces)
- Clear communication protocols (documented decisions, explicit expectations)
- Sensory accommodations (noise-cancelling headphones, lighting adjustments)
- Regular, direct feedback
- Explicit rather than implicit social rules
One ML team lead I worked with in Manchester had a brilliant approach. He created a team operating manual that explicitly documented all the unspoken social rules and expectations, when to Slack vs email, how to indicate you're deep in focus, how decisions get made. This helped everyone, not just neurodivergent team members.
That's the thing about these adaptations, they typically improve the experience for all candidates and employees, not just neurodivergent ones. Clear communication, multiple assessment options, explicit expectations, who doesn't benefit from these things?
There's a question I always ask hiring managers when we talk about this topic: are you hiring for social conformity or technical ability? Because if it's the latter, you need a process that doesn't filter candidates based on the former.
As competition for ML engineering talent continues to intensify through 2026, the companies that adapt their hiring to welcome cognitive diversity will have access to a wider talent pool than their competitors.
That's not just good ethics. It's good business.
Priya Sharma is an ML engineer with eight years of experience building production models across UK scale-ups and previously in Melbourne. She currently works on LLM infrastructure while writing about ML engineering careers and team dynamics. You can find resources on neurodiversity in technical hiring at the CIPD neurodiversity hub and explore video-based technical assessment alternatives at The OHub's recruitment platform.
