Hire for the platform your automations will actually run on, and make every finalist build something in a sandbox before you take their resume seriously. In the 137 AI and automation job posts we tracked between 3 May and 5 October 2026, n8n was named in 26% and Python in 20%. Zapier was in 7%.
That gap matters a fair bit in the Philippines. My read, from the profiles I screen, is that a lot of the automation experience there is Zapier and GoHighLevel work done inside a virtual assistant role. That work is useful, and it's a different job from the one most of those US posts describe.
The job in the US posts is mostly building
Building things from scratch came up in 80% of those 137 posts and automation in 72%. AI agents or LLMs were in 53%, integrations in 47%, and APIs or webhooks in 24%. CRM admin was 8%. Make showed up in 15% of posts, roughly level with HubSpot and Salesforce at 16% each.
So the person most of these companies want can read an API doc, catch a webhook, parse whatever JSON comes back and decide what happens when it comes back wrong. Some of them also want that person wiring a model into the middle of the flow. Treat those shares as floors, since a post only counts when the term appears in its title, summary or extracted stack.
Search the Philippine boards for "automation" and both kinds of candidate turn up in the same results. Someone strong from the VA side will keep a dozen Zaps alive and probably know your CRM better than your sales team, while a strong engineer will want half those Zaps moved into n8n with logging on every run. Decide which one you need before you post, because the ad reads the same to both of them.
What the job boards actually give you
OnlineJobs.ph and Upwork are the two boards I see buyers mention most, and they hand you very different things.
Upwork's page for Zapier developers in the Philippines is set up around projects: you post a job, invite freelancers and "only pay for approved work, hourly or by milestone." It says you may get proposals within 24 hours. The track records on that page are profile claims ("over 500+ automations deployed") that you can't check from the outside. For a one-off build that's enough, though I wouldn't hire someone to run my stack for two years on it.
OnlineJobs.ph posts read more like direct full-time hiring. One full-time automation specialist post there is a decent model of a careful process. It asks for advanced Make and Zapier with error handling, webhooks and JSON, calls the portfolio "the most important part" of the application, and then runs a writing test and a paid performance test on a dummy account with sample data. It says the whole thing takes 4 to 14 days. It also says, in capitals, that the job "MUST be your only contract." Whatever the reason that line got written, copy it. Ask every finalist about other contracts and put exclusivity in the agreement before day one.
There are plenty of people to screen. IBPAP, the industry's own body, said the Philippine IT and business process sector passed $40 billion in revenue in 2025 with 1.9 million digital workers. That figure isn't broken down by role, so it says nothing about how many of them have shipped an n8n workflow against a live API. Screen hard anyway.
Vet on a build in your own workspace
A portfolio screenshot of a canvas doesn't tell you whether anything ever ran. A paid test does. I give finalists a sandbox account in the tool they'll actually use and a task built to break.
- A webhook that receives a new lead with a messy payload, say a missing email and a nested company object.
- Push it to a test CRM, dedupe against existing records, and write the result back.
- Make one step fail on purpose and see whether anyone gets told.
- Ask for a short written note on what they built and what would break first at ten times the volume.
Then spend an hour on why they built it that way. Strong candidates bring up rate limits and error routes before you ask, and I've learned to worry about anyone who shows me a green run and stops talking. The writing test matters too, because the note they hand you is pretty much the documentation you'll get for the next year.
Build the test in an account you own, and keep it that way when they start. Make, for one, won't let you change who created a scenario, which I go into in the guide on hiring a full-time Make expert. Give the hire their own seat in your workspace on day one, so nothing they build lives in an account you'd have to ask for back.
When to hire in the US instead
The 25 automation posts that printed a salary had a median midpoint of $135,000 base. The BLS says private employers spent $54.06 an hour on wages and $24.82 on benefits for management and professional staff in June 2026, about 46% on top of pay. That puts a US automation engineer at roughly $197,000 a year, or about $16,400 a month. Twenty-five posts is a small sample, so treat the salary as directional.
Pay it anyway if the automations touch regulated data and your security team won't approve a contractor outside the US. That settles it. Pay it too if most of the work is sitting with your finance or ops leads working out what the process even is, since that needs someone in the room more than someone who builds fast.
And if you only have two or three workflows that rarely change, you don't need a full-time engineer from anywhere. A freelancer on a milestone contract is the cheaper answer.
For everything else, the build-and-run work most of those 137 posts describe, we place AI and automation engineers from the Philippines full-time on your team, in your tools, from $4K a month all-in, with a free replacement if the fit is wrong.
Write the broken-webhook test before you write the ad. If you can't describe what failure looks like in your own stack, the candidate can't either, and you'll end up hiring whoever has the prettiest screenshots.