Prompt Engineer Jobs: Skills, Salaries and Hiring Sites

You've probably seen it already. A team in Mexico City, São Paulo, or Medellín wants someone who “works with AI,” but the title shifts between prompt engineer, AI engineer, automation specialist, and something vague enough to hide the underlying work. That's the trap: if you search only for prompt engineer jobs, you'll miss most of the openings where prompt work is embedded.

The market moved fast, but the title didn't become a clean career ladder. In the U.S., postings explicitly mentioning prompt engineering jumped from about 1,400 in 2023 to nearly 6,300 in 2024, while generative-AI mentions rose from about 16,000 to more than 66,000 and large-language-model mentions from about 5,000 to 20,000Lightcast via Stanford AI Index 2025. That's the right way to read this niche, prompt engineering became a recognizable skill inside broader AI hiring, not a separate lane with thousands of cleanly labeled openings.

What Prompt Engineer Jobs Look Like in 2026

A prompt engineer title still exists, but it is narrow. A 2025 analysis of 20,662 LinkedIn job postings found only 72 explicit prompt-engineer roles, less than 0.5% of the sample academic analysis. The practical reading is simple, the work is real, but it usually shows up inside broader AI hiring, not as a clean standalone path.

The job sits inside other teams

In practice, prompt work shows up inside product teams, ML groups, AI engineering squads, and automation roles. Employers are not looking for someone who only writes clever instructions for a chatbot. They want someone who can test outputs, compare versions, detect failure modes, and work with developers or product managers to ship something stable.

For LATAM candidates, that changes the search strategy. If you are in Buenos Aires, Lima, Bogotá, Santiago, Mexico City, or São Paulo, do not treat “prompt engineer” as the only keyword that matters. Search for adjacent roles where prompt development is part of the workflow, then read the description for evaluation, RAG, versioning, APIs, and deployment language.

If you want a quick reality check, compare public AI hiring pages with the way companies frame these roles. HiredBySkill's OpenAI job listings is a useful example of how the title appears alongside broader AI responsibilities rather than as a standalone, isolated function.

Practical rule: if a posting talks about prompts but also asks for testing, Python, API integration, or collaboration across functions, it is a real prompt-adjacent role. If it only asks for “creative prompt writing,” it is usually not the role you want.

Read the title as a signal, not the job

The strongest candidates do not apply to the title. They apply to the work. That is the difference between searching for a label and understanding what the hiring manager needs.

On LATOjobs, that means filtering by AI, software, data, and automation categories, then scanning for prompt responsibilities hidden inside broader job families.

Core Skills Employers Screen in AI Talent

A prompt engineer title does not stand alone in hiring. Employers screen for people who can make model behavior reliable in production, and they usually read that through a broader AI engineering lens. The job postings analysis cited earlier shows the mix clearly, with AI knowledge in 22.8% of roles, communication in 21.9%, prompt-design ability in 18.7%, and creative problem-solving in 15.8%academic analysis. Earlier job-ad research also kept pointing to NLP, machine learning, and large language models as recurring requirements DevDiscourse summary.

Model understanding comes first

This is the first filter. Hiring teams want people who understand LLM behavior, evaluation, hallucination handling, and grounding. Show that with an evaluation harness, prompt comparison notes, and a short write-up that explains failure cases. A gallery of clever prompts does not carry much weight.

Engineering is the filter that removes hype

Python matters because prompt work quickly turns into scripts, API calls, test runners, and integrations. Knowledge of LangChain, LlamaIndex, and DSPy shows you can work inside modern AI stacks, not just chat interfaces. A clean GitHub repo, a reproducible notebook, and a small API-based prototype tell recruiters more than a polished slide deck.

Hireability test: if you cannot show how your prompt changed system output, cost, latency, or error rate, you are still describing a hobby, not a job skill.

Collaboration and domain fluency close the loop

Communication is the other half. Teams need people who can explain trade-offs to product, engineering, and business stakeholders without overselling the model. Domain fluency matters too, especially for LATAM candidates presenting work in English, Spanish, or Portuguese, because remote employers want someone who can handle multilingual workflows and sector context.

Skill clusterExample artifactsShare of postingsModel understandingEvaluation harness, failure-mode notes, grounding tests22.8%Prompt engineeringRAG workflow, few-shot examples, versioned prompts18.7%Technical stackPython scripts, API prototype, vector search demo9.2% Python, 10.8% DevOpsBusiness acumenROI memo, product brief, experiment summary15.8% creative problem-solving, 21.9% communication

The cleanest way to prove this skill mix is to document real work as you build it. A guide to hands-free note capture helps if you want to keep better records of prompt tests while switching between tasks.

For job searches, Use the LATOjobs job search resources to compare how these signals show up across roles instead of guessing from the title alone.

Salary Ranges in USD Across LATAM Hubs

Prompt engineer pay in LATAM needs a tighter lens than “it depends.” For international remote work, current market estimates put prompt-engineering compensation around $29 to $40 per hour, which translates to roughly $60,000 to $83,000 annually for full-time work LATOjobs salary context. U.S. compensation sits much higher, with one benchmark placing it around $133,000 to $191,000. That gap is why employers often treat prompt work as an embedded AI engineering skill, not a standalone title.

Use city-based ranges, not vague regional guesses

The right benchmark changes with the hiring setup. Local payroll, USD contracting, and direct remote employment are different offers, and mixing them leads to bad negotiations. Buenos Aires, São Paulo, Mexico City, Bogotá, Santiago, and Lima all sit in the same broad remote talent pool, but the stronger signal is whether the company is paying for nearshore support or full international output.

CityRole focusAnnual USD rangeBuenos AiresPrompt-adjacent AI roles in startups or agency teamsAround the LATAM remote bandSão PauloAI engineering, automation, and product-facing AI workAround the LATAM remote bandMexico CityAI engineer or data-heavy prompt workflowsAround the LATAM remote bandBogotáConversational AI, support automation, and LLM opsAround the LATAM remote bandSantiagoAI product or ML-adjacent rolesAround the LATAM remote bandLimaRemote contractor work for North American teamsAround the LATAM remote band

That band is the honest anchor. Cities with stronger access to international tech teams often push toward the top end, while local payroll roles and junior prompt-adjacent work sit lower. The point is not to pretend each city has a separate benchmark when the actual variable is the employer's pay model.

Know what you're negotiating against

For local employment, offers often follow domestic budgets and payroll rules. For USD contracting, the true comparison is the company's international hiring budget, not a local posting from your city. If you are in Argentina, Brazil, Colombia, Mexico, Chile, or Peru, do not accept local pricing for work that clearly supports distributed AI teams.

A clean negotiation starts in USD, then gets backed by evidence of shipped work. That keeps the conversation out of vague “market rate” talk and forces employers to price the role for what it is.

How to Search LATOjobs for the Right Openings

The fastest path is to stop searching for the title first. Start with remote AI and engineering filters, then layer in location and role family. On LatoJobs, a candidate in Medellín can filter remote roles, scan AI engineer or data roles, and find prompt responsibilities buried inside the description instead of chasing one keyword.

Search like a recruiter, not like a hopeful applicant

Use country browsing when you want local context, then switch to remote when you want USD exposure. A smart search sequence looks like this, Argentina, Brazil, Mexico, Colombia, Chile, and Peru for regional roles, then remote AI and software categories for distributed work. If the description mentions prompt lifecycle management, evaluation, versioning, or systematic optimization, you're in the right territory even if the title never says prompt engineer.

A candidate in Medellín might start with remote AI engineer roles, then notice the posting asks for evaluation scripts, multilingual prompts, and Python-based workflow support. That's not a stretch role, it's a prompt-adjacent opening with real production needs. The application should reflect that directly, with one project that proves you can test outputs and one resume bullet that shows measurable workflow impact.

Simple rule: if the JD hides prompt work inside AI engineer or automation language, treat the title as noise and the requirements as signal.

The same applies in reverse for employers. A lot of hiring teams think they need a prompt engineer when they need an AI engineer who can operationalize prompt behavior inside a product or support workflow.

Use the site to narrow, then read for depth

The point of the search is not to collect dozens of tabs. It's to identify postings that reward applied AI work, then tailor fast. That means one filtered search for remote roles, one country-specific search for cities like Mexico City or São Paulo, and one pass through the description to mark up the genuine skill signals before you apply.

Building a Portfolio That Lands Interviews

A portfolio for these roles needs proof, not decoration. If you show only prompt examples, you'll look like someone who uses chat tools for writing help. If you show a retrieval-augmented workflow, versioned prompts, and an evaluation harness, you look like someone who can support production AI.

Build around one shipped-style project

Start with a single project that resembles a real business use case. A clean RAG workflow beats ten generic prompts because it shows retrieval, grounding, and system behavior. Version-control the prompts, write down the model settings, and show how you tested accuracy, latency, and failure modes.

Make the evidence easy to review

Recruiters move fast. Give them a code repository, a short case study, and a summary of what changed when you adjusted the prompt or model configuration. If your project uses multilingual inputs, say so. If you handled privacy or hallucination risk, document the guardrails you added.

Your resume should match the portfolio, not sit beside it as a separate story. That means using the same language in English, Spanish, or Portuguese when you apply to different employer segments, and keeping the evidence concrete. If the project improved a workflow, explain how. If it reduced manual work, show the before-and-after process.

Use the LATOjobs blog to compare how other candidates frame their AI experience, then make your own evidence sharper than theirs.

Adjacent Titles That Open the Same Doors

The smartest search expands the funnel without weakening the target. Prompt work often sits inside other titles, and those titles are easier to find, easier to defend, and more likely to survive hiring budget changes. That's why you should map the skill overlap instead of obsessing over the label.

Adjacent titleWhere it overlapsWhat to showAI EngineerModel integration, prompt workflows, evaluationRAG workflow, API demo, test harnessML EngineerDeployment pipelines, monitoring, iterationPython scripts, metrics, version controlConversational-AI SpecialistChatbot design, NLU, user flowDialogue samples, intent tests, multilingual supportAI Product EngineerFeature implementation, product senseProduct brief, shipped feature, stakeholder notesAutomation EngineerWorkflow scripting, task orchestrationAutomation demo, error handling, logs

Pick the title that fits your stack

If you already write Python and work with APIs, lead with AI Engineer. If your strength is product thinking and feature delivery, AI Product Engineer is the cleaner route. If you've shipped internal bots or support flows, Conversational-AI Specialist is closer to how hiring managers will read your background.

For candidates in Guadalajara, Buenos Aires, São Paulo, or Bogotá, this matters because the market often rewards adjacent titles faster than a narrow prompt label. The work is the same enough to transfer, but the hiring pipeline is broader and easier to enter.

Direct advice: don't chase the prettiest title. Chase the title that matches the evidence you can defend in an interview.

A Realistic 90-Day Plan to Land Your First Role

The fastest way to miss these jobs is to spend all your time writing prompts and none of it shipping proof. Employers want evidence that you can test, evaluate, communicate, and support production work. Prompt writing alone doesn't clear that bar.

30 days to build the base

Audit your current skills, pick one domain, and ship one production-style evaluation harness. Keep the project small enough to finish, but real enough to explain in an interview. If you need help shaping your resume around that evidence, a guide to tailoring your resume is a practical reference before you start sending applications.

60 days to target the right openings

Apply to LATOjobs listings that match your stack, not just your preferred title. Tailor each application to the role language, and make sure your bullets show the same outcomes the posting asks for. Use the city filters if you want local roles in Buenos Aires, São Paulo, Mexico City, Bogotá, Santiago, or Lima, and switch to remote when you want USD work.

90 days to pass interviews

Prepare for mock evaluations, not just casual chat about prompts. Be ready to explain how you handled grounding, latency, cost, privacy, and model drift. Check references, tighten your portfolio, and keep a short list of results you can say out loud without sounding rehearsed.

Bookmark the AI category on LATOjobs, filter by your city or remote preference, and treat every application like a test of fit. The roles are real, but the bar is higher than the title suggests. Build for the work, not the buzz.

LatoJobs gives you a direct way to find prompt-adjacent AI roles across Latin America and remote USD opportunities without wasting time on vague listings. If you're serious about landing the right fit in Buenos Aires, São Paulo, Mexico City, Bogotá, Santiago, or Lima, start with the filters and work outward from there. Visit LatoJobs and use it to track roles that match the evidence you can show.