Hi {{ subscriber.first_name | default: "there" }},

This week's list surprised me.

Two roles involve autonomous driving — reviewing sensor feeds and labeling real driving scenes. One asks you to test banking AI with your own credit card. Another wants you to hum melodies with precise pitch control to train audio models. And one is paying $130–$170/hr for UK-based data scientists with big tech backgrounds.

The pattern is clear: AI companies are done with synthetic benchmarks. They want people who know how things work in the real world — on the road, in a bank, in a sales pipeline, in a recording studio — and can teach that knowledge to models.

AllJobs.work now has 70,000+ remote jobs, updated daily.

Here are the roles that stood out most this week:

🧠 Domain Expertise & Real-World Workflows

Simulate your real job
Sales Consultant — Mercor
📍 Remote (Worldwide) | 💰 $50–$90/hr
Why It's Interesting: This is not a sales job. It is a simulation design job. You build a realistic digital workspace — Drive folders, account plans, pitch decks, email threads, pricing models, Slack channels — that mirrors how enterprise sales actually works. Then you design multi-step tasks that would challenge an AI agent trying to navigate it. If you have done enterprise sales, your messy real-world workflow is the product. You are essentially building the obstacle course that the next generation of AI agents has to pass.
Apply →

Test AI with your own bank
AI Banking Workflow Tester — Sydneys Secret
📍 Remote (US and Germany) | 💰 $45/label
Why It's Interesting: This is genuinely unusual. You run small test transactions on your own credit card, then compare what actually happened against what the AI predicted — approvals, declines, cash flow logic, spending patterns. You are also reviewing AI answers to common banking questions and flagging where they break. The reason this pays well is obvious: it requires someone with a real, active bank account willing to ground-truth AI against actual financial data. That is a trust barrier most companies cannot automate around.
Apply →

Graduate-level math + Python
Research Quality Specialist - Computational Math — Turing
📍 Remote (Worldwide)
Why It's Interesting: You are not solving someone else's math problems — you are creating original, research-grade problems with scientific coding in Python, then rigorously solving them yourself. This is the kind of role that exists because AI models are getting good enough at math that the benchmarks need to get harder. If you have a graduate-level quantitative background, your ability to design hard problems is now a sellable skill.
Apply →

🚗 Autonomous Driving

AV meets language precision
Driving Scene Caption Analyst — Turing
📍 Remote (Worldwide)
Why It's Interesting: You review short sensor feeds from vehicles, evaluate LLM-generated reasoning about what is happening on the road, and refine the outputs for accuracy. This sits at the intersection of autonomous driving, spatial reasoning, and language — and that combination is rare. If you understand US road rules and can critically assess whether an AI correctly interpreted a driving scene, you are working on one of the highest-stakes applications of AI that exists.
Apply →

AV labeling, work from home
SME-AV-Label & Annotation — Cognizant
📍 Remote (US)
Why It's Interesting: Cognizant is a large tech services company, and this is autonomous vehicle labeling and annotation — the foundational work that self-driving systems are built on. The listing is light on specifics, but AV annotation is one of the more stable and well-funded corners of the AI data market. If you have 2–4 years of annotation or Hi-Tech domain experience and want something steady, this is worth a look. The fact that it is fully remote and day-shift only is a practical bonus.
Apply →

🎙️ Audio & Voice

Musical precision for AI
Vocal Audio Specialist - AI Trainer — Invisible
📍 Remote (Worldwide) | 💰 $6–$65/hr
Why It's Interesting: This is one of the most specific roles I have seen. You hum melodies with precise pitch control — major thirds, staying in key, flat tones with no vibrato — and record them with professional audio quality. The wide pay range ($6–$65) almost certainly reflects skill level: if you have real vocal training and proper recording equipment, you are at the top end. AI music and audio generation models need clean, controlled vocal data, and people who can deliver it reliably are a small pool.
Apply →

Straightforward voice work
Spanish Voice Recording Contributor — Lionbridge
📍 Remote (Worldwide) | 💰 $13–$15/hr
Why It's Interesting: European Spanish specifically. You record short sentences, transcribe audio clips, and help organize text data. It is simple, flexible, and part of the infrastructure that powers Spanish-language voice AI, transcription, and speech products. Not the highest-paying role in this list, but it is accessible, requires no AI experience, and lets you pick the tasks you prefer. Solid side income if you are a native European Spanish speaker.
Apply →

🌍 Language & Localization

MSA at a top AI lab
Data Annotation Specialist, Modern Standard Arabic (MSA) — Cohere
📍 Remote (Canada) | 💰 30 CAD/hr
Why It's Interesting: Cohere is one of the most respected AI companies, and this role goes beyond labeling. You are ranking model outputs, auditing accuracy across text and structured data, and recommending optimizations — closer to product quality work than annotation. MSA is the formal Arabic that spans the entire Arab world, and getting it right in AI models has massive downstream value. Having Cohere on your resume also opens doors.
Apply →

Dialect-specific — rarer
Data Annotation Specialist, Arabic (Najdi/Hijazi Dialect) — Cohere
📍 Remote (Canada) | 💰 30 CAD/hr
Why It's Interesting: This is the companion to the MSA role above, but dialect-specific — and that matters. Most AI models handle formal Arabic passably but fall apart on regional dialects. Najdi and Hijazi are spoken across Saudi Arabia, and the fact that Cohere is hiring separately for these dialects signals they are serious about getting Arabic right at every level, not just the formal one. If you speak these dialects natively, your competition is extremely thin.
Apply →

Cultural + linguistic QA
Language Alignment & Resource Partner (Taiwanese) — Invisible
📍 Remote (Worldwide) | 💰 $6–$65/hr
Why It's Interesting: This is more than annotation — you are the cultural quality gate. You review AI outputs for grammatical accuracy and naturalness in Taiwanese, catch subtle cultural errors, and develop educational resources so that future outputs improve. That is a feedback loop role, not a one-off labeling task. The wide pay range rewards depth: if you can autonomously identify patterns in where AI gets Taiwanese wrong and build documentation to fix it, you are at the top end.
Apply →

Hindi content sourcing
Hindi Digital Content Data Collection — OpenTrain
📍 Remote (India) | 💰 $1.40/label
Why It's Interesting: The pay is per-label rather than hourly, so your effective rate depends on speed. The work itself is sourcing real, naturally occurring Hindi digital content — emails, messages, notes, media — across everyday topics. The key requirement is that nothing can be synthetic or AI-generated. AI models need authentic human data to learn from, and if you can efficiently collect and redact real-world Hindi content, this is steady sourcing work.
Apply →

Arabic content sourcing
Arabic Digital Content Data Collection — OpenTrain
📍 Remote (Worldwide) | 💰 $1.40/label
Why It's Interesting: Same structure as the Hindi role above — collecting real, non-synthetic Arabic digital content across platforms. The fact that this one is open worldwide (not just one country) reflects how broadly Arabic is spoken and how much variety AI companies need in their training data. If you interact in Arabic across messaging, email, and social platforms daily, your normal digital life is essentially the raw material.
Apply →

🚀 Entry-Level / Getting In

No experience required
AI Training Generalist — Invisible
📍 Remote (US) | 💰 $20/hr
Why It's Interesting: The listing says it plainly: no prior experience needed. You interact with AI systems, review outputs, follow task guidelines, and help create training data. $20/hr is a reasonable entry-level rate for US-based work, and the tasks are well-defined with guidance provided. If you have been looking at AI jobs but feeling underqualified, this is designed for you. Get in, learn the workflow, and build from there.
Apply →

India-based entry point
AI Annotator — Toloka Annotators
📍 Remote (India) | 💰 $9/hr
Why It's Interesting: Rating AI content, evaluating factual accuracy, comparing model responses — this is the core loop of AI data work. $9/hr is competitive for India-based remote work, and Toloka is one of the larger platforms in this space. The value here is not just the paycheck — it is learning how AI evaluation actually works. Guidelines-based judgment, consistency, structured feedback. Those skills transfer to higher-paying roles later.
Apply →