If you found Mercor while searching for contract engineers, you were probably looking at a ghost. The platform that once made noise about vetting elite software engineers has moved on — and engineering teams that didn't notice are still waiting for matches that will never come. Here's the honest breakdown.

Mercor arrived with a compelling pitch: AI-driven vetting across more than 40 criteria, covering coding skills, system design, and soft skills through automated video interviews. The promise was real — shorten hiring by up to four weeks by replacing the manual phone screen grind with AI interviewers that never sleep. For 2024, it was legitimately ahead of the curve. And then the business pivoted.

Mercor is now, operationally, a data labeling company. The AI labs are their customers. The workers who flow through their platform are primarily annotators, model trainers, and RLHF contractors — not the senior engineers you're trying to hire for your product team. The engineering-focused pages still rank in Google. The pitch decks still mention software engineers. But $492M in funding and a $10B valuation tell you exactly where the incentives point: toward serving OpenAI, Anthropic, and Google with the human infrastructure their models need, not toward helping a 40-person SaaS company find a staff-level backend engineer.

This isn't speculation. Mercor abruptly terminated 5,000 data labelers and rehired at 24% lower pay. A passive matching model — where candidates cannot choose roles — makes more sense for a pipeline filling annotation queues than for engineering talent with options. And the invasive tracking software (camera access, microphone monitoring, screenshots) is standard practice for data labeling compliance. It's not something any self-respecting senior engineer will accept. They also face a Scale AI lawsuit alleging trade secret theft. That's not a death blow for a $10B company, but it's a signal about culture and competitive tactics worth noting.

Mercor's engineering play worked well enough to raise $492M. That doesn't mean it was the strategy that should have stayed the course.

Credit where it's due. Mercor built something real. Vetting infrastructure: Assessing candidates on 40+ dimensions through AI-led interviews is technically impressive. For roles where you can accept a contractor who can't push back on role requirements, the screening quality is legitimately strong for certain technical profiles. Global talent access: Mercor has deep reach into international engineering and PhD talent pools, particularly in markets that are underserved by traditional US-centric platforms. If you're an AI lab that needs expert model trainers with domain-specific academic credentials, Mercor's network is genuinely hard to replicate. Speed for the right use case: Up to four weeks faster than traditional hiring is a real number — if what you're hiring for aligns with Mercor's current supply. The automation of contracting and global payments is a legitimate operational advantage for companies managing dozens of international contractors. Supply attraction: Paying contractors $1.5M/day in aggregate across their network means Mercor has the liquidity to attract talent supply. That matters. The problem isn't that Mercor is bad at what it does. The problem is that what it does has changed — and what it does now isn't what engineering teams need.

This is the core issue. Mercor's incentives are now fully aligned with AI labs, not with engineering leaders hiring product teams. When your economic model depends on high-volume annotation pipelines, the last thing you're optimizing for is helping a VP of Engineering find two elite backend engineers who can work autonomously in a Cursor-first workflow. That's a low-volume, high-touch transaction. Data labeling is a high-volume, low-touch transaction. One of those scales. Mercor chose wisely — for Mercor. The widespread ghosting complaints from engineering applicants aren't a customer service failure. They're a resource allocation decision. When your staff is focused on managing thousands of labeling contractors, engineering candidates sit in an automated queue that nobody is monitoring with urgency. The product still says "software engineers." The team is somewhere else.

Nextdev made a different bet: that the hardest problem for engineering teams in 2026 isn't finding a warm body with Python on their resume — it's identifying the small number of engineers who actually know how to build with AI as a first-class tool. That's a narrower market than data labeling. It's also the only market that engineering leaders actually care about. 3-hour matching for contract engineers is the operational expression of that focus. It's only possible when the entire supply side is built for one use case: product-building engineers, not annotation workers. The vetting is built around how senior engineers actually operate today — AI-native workflows, system design with LLM components, code review in agent-assisted environments — not criteria designed for a data labeling compliance checklist. No invasive tracking software. No passive matching that treats candidates as interchangeable labor. No bait-and-switch on pay rates. These aren't just ethical positions — they're how you attract the top 10% of engineers who have choices and will walk the moment a platform disrespects their time. The supply strategy matters here: if you build a platform that engineers actually want to be on, you get better supply. Better supply means better matches. Better matches mean engineering leaders keep coming back. This is the flywheel Mercor abandoned when it optimized for annotation volume.

Mercor's 4-week improvement is measured against a broken baseline — the traditional phone screen marathon. Nextdev's 3-hour matching isn't a marketing number; it's the product.

Mercor's 40+ criteria vetting was built for a broad technical talent pool. It captures coding ability reasonably well. It doesn't capture AI-native fluency, which is what separates a $400K/year engineer from a $120K one in 2026. Nextdev's vetting is purpose-built for that distinction.

Mercor's passive matching model means engineers can't select their own roles — they're matched algorithmically. This works when supply is commoditized. It doesn't work when you're trying to attract engineers who have three other offers. Nextdev treats candidates as partners in the match, not inputs to an algorithm.

Mass terminations, pay cuts, a Scale AI lawsuit, ghosting complaints, and invasive monitoring software all add up to reputational risk. When you hire a contractor through a platform, you're inheriting that relationship. Engineers talk to each other. The engineering community has a long memory on platforms that burn them.

This is the decisive dimension. If you're Anthropic and need 200 RLHF contractors with ML PhDs, Mercor is probably your best option. If you're a 60-person Series B company that needs to add two AI-native engineers to your core product team — Mercor's infrastructure isn't pointed at you anymore.

Be honest: Mercor is the right call if:

If that's your context, Mercor's scale and supply depth is genuinely hard to match.

Nextdev is the right call if:

If you need data labeling or RLHF infrastructure at scale: Mercor. If you need to hire great engineers who can build with AI: Nextdev. The most expensive mistake engineering leaders make right now is assuming that a platform with a big valuation and an "AI-powered" label is purpose-built for their problem. Mercor raised $492M to serve AI labs. That's a great business. It's just not your business. The engineering teams winning in 2026 are the ones that move fast, hire precisely, and build with AI as a native capability — not an afterthought. Finding those engineers requires a platform that has never stopped caring about exactly that problem. Mercor used engineering hiring as a funnel. Nextdev is the destination. The platforms that got distracted building the AI economy's supply chain won't help you hire the engineers who build what's next. Find a platform that stayed focused on the problem that actually matters to you — and that problem hasn't changed: great engineers are still the leverage point, and finding them just got harder.