Role Description We are sharing a specialised part-time consulting opportunity for experienced machine learning engineers and researchers with strong practical expertise in ML research workflows, model training, post-training, dataset curation, reinforcement learning, architecture design, evaluation tasks, and sandboxed technical environments. This role supports current and upcoming remote consulting opportunities focused on benchmarking AI agent performance on realistic machine learning research tasks. Selected professionals will complete self-contained ML research tasks under defined time and compute constraints, provide human reference performance for evaluation workflows, and submit structured work outputs that support technical assessment and research-quality analysis. Key Responsibilities - Machine Learning Research Task Execution - Attempt open-ended machine learning research tasks under fixed time and compute constraints. - Work independently in a sandboxed Linux environment using provided compute resources. - Apply practical ML engineering and research judgment to self-contained AI R&D tasks. - Use preferred development workflows and tools, including IDEs, coding assistants, notebooks, or command-line workflows where permitted. - Submit final work products that reflect clear reasoning, technical execution, and reproducible effort. - Benchmarking & Human Reference Evaluation - Serve as a skilled human reference point for evaluating AI agent performance on realistic ML research tasks. - Complete tasks using the same constraints and environment conditions defined for evaluation workflows. - Support benchmark quality by producing reliable, interpretable, and technically meaningful task attempts. - Document decisions, assumptions, implementation choices, and constraints where relevant. - Complete short pre-task and post-task questionnaires as part of the evaluation process. - Technical Workflow, Recording & Quality Control - Work in sandboxed technical environments with Linux, internet access, and provided compute resources. - Record full working sessions when required for evaluation and review purposes. - Follow task-specific confidentiality, NDA, environment, and submission requirements. - Debug issues involving code, packages, model training workflows, data processing, runtime behavior, or environment setup. - Submit required materials, including final outputs, recordings, questionnaires, and supporting notes where applicable. Qualifications - 3+ years of practical machine learning experience, with time spent in a PhD program counting toward this requirement where relevant. - Hands-on experience with at least one major ML framework such as PyTorch, JAX, or TensorFlow. - Strong practical ability to complete open-ended ML research or engineering tasks independently. - Experience working in Linux environments, debugging technical workflows, and managing research-oriented development tasks. - Ability to reason under time and compute constraints while producing clear, high-quality technical work. - Strong written communication skills for explaining methods, decisions, and results. - High attention to detail and comfort following structured evaluation, recording, and submission requirements. - Availability for at least 20 hours per week if selected, with additional availability considered helpful depending on project needs. Educational Background - Strong academic or professional background in machine learning, artificial intelligence, computer science, data science, statistics, engineering, or a related technical field is highly relevant. - PhD experience, advanced research experience, or comparable industry experience in machine learning may be especially valuable. - Candidates with experience from highly selective academic programs, major technology companies, AI research teams, or comparable technical environments may be a strong fit. - Practical research and engineering experience may be considered alongside formal education depending on project requirements. Nice to Have - Deep hands-on expertise in one or more of the following areas: - Pretraining transformer language models from scratch. - Reinforcement learning, PPO, reward shaping, custom gym or gymnasium environments, and throughput tuning. - Full fine-tuning, LoRA, QLoRA, DPO, RLHF, RLAIF, distillation, or post-training workflows. - Large-scale corpus filtering, deduplication, subsampling, and benchmark contamination avoidance. - Architecture design under strict parameter-count or size constraints. - Modifying pretrained architectures, including attention patterns, pooling heads, or training objectives. - Contrastive training for embedding or retrieval models. - Generative vision or video modeling. - Multilingual or low-resource language work. - Image or video data pipelines at scale. - Balancing competing model objectives such as safety and capability. - Prior experience as an ML evaluator, red-teamer, benchmark contributor, research engineer, or technical baseliner. - Experience using AI coding assistants in technical workflows while maintaining strong independent judgment. - Comfort working with confidential project materials and structured technical review processes. Why This Opportunity - Apply advanced ML research and engineering expertise to realistic AI R&D benchmarking tasks. - Serve as a human reference point for evaluating AI agent performance on open-ended technical challenges. - Work with provided compute and sandboxed environments without requiring personal GPU resources. - Use practical ML judgment across training, post-training, dataset curation, architecture, reinforcement learning, or evaluation workflows. - Remote structure with competitive hourly compensation. Contract Details - Independent contractor role. - Fully remote with project-based technical work. - Minimum expected availability of approximately 20 hours per week if selected, with greater availability preferred depending on project needs. - Competitive rates of $60–$75 per hour depending on expertise, ML research depth, task performance, and project scope. - A work-trial-style baseline task may be required before longer-term selection. - Each assigned baseline task may be attempted only once per contractor. - Project work may require screen recording, questionnaires, final work product submission, and adherence to confidentiality or NDA requirements. - Compute and sandboxed technical environments may be provided depending on task scope. - Weekly payments via Stripe or Wise. - Projects may be extended, shortened, or adjusted depending on scope and performance. About the Platform This opportunity is available through 24-MAG LLC. We connect experienced professionals with remote consulting opportunities across technical, evaluation, and project-based workstreams.