Yes. Health AI training and evaluation work is carried out remotely, in your own time, at times of your choosing.

Do I need technical or coding skills?

No. Clinical judgement and real-world experience are what is required. Most tasks involve reading, reasoning, and writing assessing AI outputs, defining reference answers, and quality checking responses.

Is patient data involved?

No identifiable patient data is used.

Is this a job or freelance contract work?

This is contract project work, not employment. Most arrangements are flexible and task-based rather than salaried. You will be contracting directly with Mercor or Micro1, not with this site.

Can I do this alongside NHS or locum work?

In many cases yes subject to your employer's policy on secondary work and any conflicts of interest.

Optimising Your LinkedIn for AI Training Work

By Sean Key, Editor — Applied Clinical Judgement. Last reviewed July 2026.

On these platforms your LinkedIn is not just a formality you paste into a box. It is a matching signal the system reads, and on Mercor a strong LinkedIn presence can trigger invitations to you directly. So it is worth ten minutes to make sure yours is working for you rather than sitting half-finished.

This page covers what to fix, in priority order, and the one rule that ties your LinkedIn to the rest of your application.

Why it matters here

Both Mercor and Micro1 ask for your LinkedIn as part of your profile, and both use your professional signals to match you to work. A complete, specific profile gives the matching system more to work with; a sparse one gives it less, and you surface for fewer opportunities. Think of it as a second parseable source of truth about you, alongside your CV.

It is worth understanding how this actually works, because it is not magic. Matching systems and human recruiters alike find you by searching for terms – the skills, tools, and role words that describe what you do. If the words a recruiter or an algorithm searches for are not somewhere in your headline, your summary, your job titles, or your skills, you simply do not come up. Your experience could be perfect and it would not matter, because you were never found. So the job of your profile is first to be findable on the right terms, and then to be convincing once someone lands on it.

When someone does land, they decide fast. In the first few seconds they are really answering three questions: what do you do, are you credible, and do you fit. Your headline answers the first, your quantified experience answers the second, and your skills answer the third. If a profile does not answer all three quickly, the reader moves on – so every section below is really about answering one of those three questions without making the reader hunt for it.

What to fix, in order

Headline: say what you do, specifically. “Consultant” or “Analyst” is too thin. “Equity research analyst – financials and energy” or “Physiotherapist – musculoskeletal and rehab” gives the system real terms to match. Use the words that describe your actual expertise.
About section: your core story in plain terms. A few lines on your domain, your depth, and the kinds of problems you handle. Specific beats impressive, and concise beats long – clear, economical writing is itself part of what this work rewards, so a tight summary reads better to both the algorithm and the person. This is prose the matching signal can read.
Experience: mirror your CV, then quantify it. Same roles, same titles, same dates – and every role carrying numbers, covered in the next section because it matters most.
Skills: list a lot of real ones and pin the right three. Covered in detail below, because the count and the pinning both matter.
Everything else: fill it all in. Certifications, qualifications, publications, projects, languages, endorsements, recommendations – all counted signals, all covered below.
Keep it current. An out-of-date profile matches you to the wrong things or nothing. Update it when your situation changes.

Write for an algorithm that knows nothing about your job

This is the half most people miss, and it is where the real score is won. The system reading your profile has no idea what your job involves. It does not know that “led the migration” was a two-year programme across 37 sites, or that “improved reporting” saved a team three days a month. If you do not spell it out in plain terms with numbers, that impact is invisible – to the algorithm and to the human skimming behind it.

So write every role as if explaining it to a smart outsider who knows nothing about your field. The structure that works is a situation, the task, the action you took, and the result – but weighted heavily towards the result, and the result in numbers. Not “responsible for improving performance” but “cut report turnaround from 5 days to 1, across a team of 12”. Put a number in every role you can: how many people, how much money, what percentage, what scale, over what time. Revenue, users, sites, hours saved, team size, error rates – whatever your field measures. Vague competence reads as filler; a number reads as proof, and proof is what gets you found and believed.

Two rules make this land. Spell out the acronyms and context an insider would assume, because the algorithm is the outsider. And put the number first where you can, so it is the thing that gets read before the reader moves on.

List everything the algorithm counts

The profile score is built from completeness signals, and the system literally counts them. A profile that lists only roles and a few skills is leaving easy points on the table. Fill in every section that applies to you:

Skills: list 15 to 25 real ones. This is not the place to be modest. Include technical skills, tools, and domain expertise – the specific terms, not broad categories. Then pin your top three to match the work you are targeting, because those three are the most visible and are used directly as search filters.
Certifications and qualifications: add all of them. Every industry certification, professional qualification, and credential you hold. They validate your expertise to a system that cannot otherwise verify it, and they are a counted signal.
Publications, projects, and portfolio pieces. Anything you have authored, built, or delivered – papers, case studies, named projects, portfolio work. These are proof of what you have actually done, and they carry weight in the score.
Languages: list every one you can work in professionally. Not just your first language – every language you could genuinely do the work in. Some AI training work specifically needs particular languages, so this can be the exact thing that matches you to a project, and it is a signal the algorithm reads directly.

Collect endorsements and recommendations

These are the social-proof signals, and the algorithm measures them too – so they are worth actively gathering, not waiting for.

Endorsements sit against your skills. The more of your listed skills carry endorsements, the stronger those skills read. A practical way to build them: endorse colleagues genuinely, and many reciprocate.
Recommendations are the written references against your roles, and they are underrated. Even two or three genuine ones strengthen a profile noticeably. Ask former managers and colleagues directly – most are willing if you make it easy by suggesting what to focus on.

Both take a little effort to collect and both move the score, which makes them some of the highest-return work you can do on your profile.

What puts recruiters off

Some things actively cost you, and they are easy to fix once you know to look for them. Claims with no proof behind them – “motivated”, “hardworking”, “passionate” – read as filler, because anyone can type them; a number or a concrete result does the job those words are pretending to do. A missing photo, an empty or one-line summary, and long unbroken walls of text all make a reader lose confidence fast. So do job titles that do not match how your field normally names things, because both the search and the human reader are looking for the standard term. And unexplained gaps invite questions; a one-line note on what a gap was is better than leaving it blank. None of these are hard to fix, but each one turns readers away.

The one rule: same story everywhere

This is the rule that catches people out. Your LinkedIn, your CV, and your interview answers are cross-checked. On Micro1 the system compares your spoken answers against your CV and LinkedIn to build a consistency score, and a mismatch counts against you – if your LinkedIn presents you as one kind of professional and your interview examples point somewhere else, that inconsistency shows up.

So the fix is simple to state and worth the effort: decide your core professional story, then make your CV, your LinkedIn, and the examples you will use in the interview all tell it. Same titles, same dates, same emphasis. You are not building three separate profiles; you are presenting one person, consistently, three times.

What you do not need

You do not need a huge following, daily posts, or a personal brand. These platforms are matching your expertise to work, not measuring your influence. A complete, accurate, specific profile that matches your CV does the job. Do not spend effort on reach when the return is in accuracy and completeness.

And do not let your LinkedIn hold up your application. Get the headline, About, experience, and skills to a solid, honest standard – a day’s work, not a fortnight’s – then apply, and keep improving the profile afterwards. Where places on a role go to whoever qualifies first, being in early with a good profile beats being late with a flawless one.

Frequently asked questions

Do I actually need a LinkedIn profile?

It is expected as part of your profile on both platforms, and on Mercor a strong one can prompt invitations directly. If you genuinely do not have one, the platforms let you say so, but a good profile is a real advantage worth creating.

Does my LinkedIn need to match my CV exactly?

The core facts should match – titles, dates, the story they tell. The systems cross-reference your CV, LinkedIn, and interview answers, and inconsistencies count against you. Consistency is the point, not identical wording.

Do I need lots of followers or posts?

No. This is about matching your expertise to work, not building an audience. Completeness and accuracy matter; follower count does not.

What is the single highest-value fix?

A specific headline and a specific skills list. Those are the terms the matching system reads most directly, and vague ones waste the slot.

Which keywords should I use?

The real words for what you do – the skills, tools, methods, and standard role titles someone would search for to find a person like you. Put them where they get searched: your headline, your summary, your job titles, and your skills. Use the terms that genuinely apply to you; the aim is to be findable for the work you can actually do, not to game the search.

How much do I really need to quantify?

As much as you honestly can – a number in every role is the target. The system reading your profile does not know your field, so “improved efficiency” means nothing to it, while “cut a 5-day process to 1 day for a team of 12” means everything. Numbers are what turn vague competence into proof.

Should I list languages if the work is in English?

Yes, list every language you can work in professionally. Some AI training projects specifically need particular languages, so a language you left off could be the exact thing that would have matched you to paid work. It is also a signal the profile score counts directly.

Are endorsements and recommendations worth chasing?

Yes – both are measured by the algorithm and both are underused. Endorse colleagues genuinely and many return the favour on your skills; ask former managers directly for two or three written recommendations. It is some of the highest-return effort you can put in, because most people skip it.

The bottom line

Your LinkedIn is a matching signal, so treat it as one. Make the headline and skills specific, quantify every role for a reader that knows nothing about your field, and fill in every counted section – certifications, qualifications, publications, projects, languages, endorsements, recommendations. Keep it current, and make sure it tells the same story your CV and interview will. You do not need to be an influencer – you need to be findable, complete, and backed by numbers.

Your LinkedIn is one stage of getting hired. See our full getting-hired guide for the CV, the interview, getting matched, and the rest.

—

Applied Clinical Judgement is a referral service run by Sean Key, a Digital Health Senior Programme Manager with 29 years of NHS and private-sector experience. [LinkedIn] · [Book a vouching call] · [Application guide (PDF)]. ACJ is not an employer or recruiter; it points you to the platforms that do the hiring, and may receive a referral credit at no cost to you.