Design, build, and refine MuJoCo simulation environments that train AI systems to perform real-world robotic tasks — from locomotion and dexterous manipulation to multi-agent coordination. Work includes wrangling MJCF files, tuning reward functions, and debugging contact dynamics for reinforcement learning pipelines.
Design and implement coding benchmarks used to evaluate frontier AI models across real-world programming tasks. Build and maintain scalable data pipelines for AI evaluation workflows, analyze AI-generated code for correctness and edge-case failures, and create structured evaluation scenarios that rigorously test reasoning, debugging, and code quality across large multi-language codebases.
Work across a broad range of tasks to research, evaluate, and improve AI model outputs as a flexible generalist contributor.
Play and evaluate games to generate gameplay data and feedback that helps train and improve AI systems.
Review AI-generated investment and banking outputs as a Subject Matter Expert, evaluating valuation logic (DCF/comps/precedents), deal structuring, capital markets concepts, accounting linkages, and risk analysis. Provide precise written feedback, fact-check financial claims, and write model solutions demonstrating correct methods.
Perform general AI training and evaluation tasks requiring bilingual proficiency in English and Kannada, including content review and quality assessment.
Evaluate AI-generated content for safety and quality in both English and Urdu, identifying harmful outputs and improving model behavior.
Review AI-generated accounting and finance responses as a Subject Matter Expert, evaluating bookkeeping logic, journal entries, accruals, reconciliations, financial statement presentation, and controls. Fact-check technical accounting claims and write model solutions demonstrating correct methods.
Review and audit physics research outputs for technical errors and quality to ensure high-accuracy AI training data.
Apply high-level scientific judgment and cross-disciplinary perspective to adjudicate physics research outputs for AI training quality.
Develop features and tools using O3DE, Python, and C++ to train AI on open-source 3D engine development.
Solve complex coding problems and review AI-generated code across multiple programming languages to improve model accuracy.
Page 1 of 54 · 639 total
PARA AI Labs aggregates active AI training jobs, freelance AI projects, LLM evaluation roles, and data annotation work from leading platforms such as xAI, TELUS Digital, Scale AI, Turing, and Mercor — all in one place, updated daily.
No more searching across 10+ websites for AI projects.
See compensation upfront before you apply.
One click takes you straight to the source platform.
Legitimate, well-paying projects from reputable companies only.