Frontend Engineer — Web Replication Preference Rater | $90/hr Remote
Overview
This project feeds a dataset of human preference judgments on AI-generated frontend code. Each unit supplies a real crawled reference site as a zipped tree with screenshots, plus two model-built replicas; you run all three on your own machine, render them at a 1920×1080 viewport, and judge which attempt is the better reproduction state by state. Grading covers visual fidelity (box model, typography, color, z-order) and how the code is constructed, not just how the render looks. The work suits a practicing web developer who can read hand-written HTML and CSS across old and modern eras, and who treats browser DevTools as muscle memory.
What You'll Do7
- 1Download the reference site and both model attempts, unzip all three, and serve them from a local static server so relative asset paths resolve.
- 2Open each page at a 1920×1080 viewport and reproduce every required state, including states that appear only after hovering, clicking, or scrolling.
- 3Compare the two attempts against the reference state by state, scoring visual fidelity across box model, spacing, typography, colors, borders, images, z-order, and overflow.
- 4Read the source of both attempts and judge construction quality, catching shortcuts like hardcoded pixel offsets, absolute-positioned layout, inline style soup, flat div trees, or pasted screenshot images.
- 5Test responsiveness by resizing the viewport and identifying where layouts break, then explain the precise mechanism, such as a nav collapsing below 1400px because of a missing media query.
- 6Write an evidence-cited justification for every preference, referencing specific source lines or DOM behavior instead of saying one attempt simply looks closer.
- 7Use the task-flagging escape hatch only for tasks that cannot be completed from the provided screenshots, and back the flag with reasons.
Requirements8
- 13-8 years of professional web development experience with a focus on frontend or full-stack work.
- 2Strong command of hand-written HTML and CSS, including semantic markup, flexbox, grid, media queries, and legacy float- and table-based layouts.
- 3Ability to read JavaScript and explain what page scripts do to the DOM, even if you do not write JS daily.
- 4Browser DevTools fluency: setting exact viewports, walking the element tree, inspecting computed styles, and observing hover-handler mutations.
- 5Command-line comfort: unzipping archives, starting a static local server, and troubleshooting broken asset references.
- 6Professional written English for the free-text justifications.
- 7A completed Mercor engagement delivered in full; this project does not onboard net-new experts to the seat.
- 8A desktop or laptop with administrator rights, capable of displaying a 1920×1080 viewport.
Who Should Apply
The ideal candidate is a practicing frontend or full-stack developer who has shipped interfaces for several years and can read a page's source to infer how it is built. Backend-only, ML/data-science-only, mobile-native-only, and DevOps-only engineers should skip this, as should designers who do not write code; neither group can grade the source axes. Framework-only developers who have never written raw CSS outside a component library will also struggle, since many reference pages are legacy table-and-image-map builds. Common rejection reasons include failing to deliver a full Mercor engagement, lacking a real local environment with admin rights, or submitting vague justifications that do not cite specific source or render evidence. Prior RLHF, preference-labeling, or structured code-review experience is the strongest single signal that a candidate will ramp quickly.
Salary Insight
The rate is $90.00 per hour for completed units. That sits at the upper end for contract frontend evaluation work and matches the expectation that you work in focused 2-3 hour blocks and deliver rubric-grade judgments, not microtask-style output.
Location
Required Skills
Application Tip
Before applying, run one of your own past frontend projects through the same lens: resize it down to 1400px and below, identify the first layout break, and write a short evidence-cited paragraph explaining the DOM and CSS cause. Mention that exercise in your application, along with how many completed Mercor engagements you have delivered.
See NearSkill jobs more often in your search
How your application is processed
1Application received
Your resume and details are logged the moment you apply.
2ATS + eligibility screening
We check your profile against the role’s skills, seniority, and requirements.
3Employer sees qualified profiles only
Only candidates who clear screening move forward.
Similar open positions
Explore active roles that match your skills and interests.
SME Careers
VerifiedHTML CSS Team Lead for Remote Frontend QA Contract
As an HTML and CSS Team Lead, you oversee quality and consistency for AI training projects guided by remote contributors. Review AI-generated HTML and CSS snippets against detailed rubrics, delivering precise feedback on semantics, accessibility, and responsive layouts. Lead conversations on ARIA basics, CSS specificity, and cross‑browser compatibility while guiding trainers through updated guidelines on Discord and shared docs. This remote, hourly contract with SME Careers focuses on frontend quality leadership for training data that must be accessible, well-structured, and aligned with client expectations.
SME Careers
VerifiedReact and Next.js Engineer for AI Frontend Review
This remote contract role focuses on reviewing React and Next.js code, frontend architectures, and AI-generated explanations. You will produce high-quality reference implementations and step-by-step reasoning to guide AI training and improve model performance. Evaluate solutions for correctness, clarity, accessibility, security, and performance across SSR/SSG/ISR patterns and routing. You’ll help shape frontend guidance used by large AI labs through SME Careers and contribute to practical frontend quality standards.
Micro1
VerifiedFrontend Engineer Specialist
A remote contractor role for a frontend engineer who excels at explaining complex frontend decisions in writing. You’ll apply deep knowledge of JavaScript, TypeScript, and React to provide clear rationales, critique code samples, and draft technical docs that guide frontend work. The position emphasizes documentation, architectural reasoning, and real-world engineering insight to help train next‑generation AI systems.
Micro1
VerifiedFrontend Specialist for AI Training Documentation and UI Delivery
A remote contract frontend role focused on AI training support. You’ll translate detailed component style guides and screenshots into pixel-perfect, responsive UI outputs that align with a formal design system. The work centers on producing high-quality technical documentation, examples, and structured outputs to help train next generation AI models, rather than building production apps. Strong communication and careful visual execution are essential as you collaborate asynchronously in a distributed team. Typescript and CSS3 drive the UI work, with accessibility and performance considerations baked in.
SME Careers
VerifiedAngular Engineer for AI Frontend Review (Remote)
Operate remotely on an hourly contract to review AI-generated Angular code, frontend architectures, and technical explanations. Produce high-quality reference implementations and step-by-step reasoning for complex engineering problems using TypeScript and Angular patterns. Assess solutions for correctness, clarity, accessibility, security, and performance, with attention to change detection, RxJS streams, and Angular Universal SSR concerns. Deliver model solutions that demonstrate best practices and compare multiple AI responses based on reasoning quality.
SME Careers
VerifiedJavaScript Engineer for AI Training Content and Evaluation
Remote, hourly contract for a JavaScript engineer focused on AI training data evaluation; you will review complex AI-generated code and explanations for accuracy and clarity. Assess reasoning quality, identify edge cases, and deliver precise feedback that helps models improve step-by-step solutions. Contribute to creating high-quality training content and participate in evaluation workflows with AI coding assistants. Candidates should bring production-level JavaScript experience and familiarity with React, Node.js, or Express, plus clear English writing to explain technical concepts.