About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. We connect experienced contributors with short-term, project-based work that directly shapes how state-of-the-art AI systems behave.

We hire and contract contributors directly. This role is part of OpenTrain’s ongoing efforts to create high-quality training data and task specifications used to teach and evaluate large language models and other AI systems.

Why AI training and data labeling work matters

AI training (also called data labeling or human feedback work) is the human side of building intelligent systems — people create, verify, and rate the examples models learn from. Contributors enjoy flexible, remote work that can be done part-time and often requires strong domain or technical skills rather than a traditional office job.

• 100% remote, flexible schedules — fit around other commitments.
• Work is accessible to specialists and domain experts; your work directly influences model behavior.
• Many projects require no prior labeling experience; specialist roles (like this one) pay more for deep technical skill.

The Role

OpenTrain is seeking senior data science experts to author and verify original, end-to-end computational data science problems and GenAI prompts. You will create reproducible, computationally intensive tasks that reflect realistic business workflows across industries such as telecom, finance, e-commerce, government, and healthcare.

This is a contract, part-time engagement during active project phases. Expect to contribute approximately 10–20 hours per week on average; work is project-based rather than permanent.

• Role type: Contractor, part-time, project-based.
• Location requirement: United States (applicants must be located in the USA).
• Pay: Up to $40/hour (typical range $15–$40/hr depending on task and expertise).

What you'll do

Design original data science problems and complete solution artifacts that are computationally non-trivial, reproducible, and grounded in realistic business contexts. Problems should span the full analytical workflow and include verification materials.

• Author end-to-end Python problems covering ingestion, cleaning, EDA, feature engineering, modeling, validation, and deployment considerations.
• Produce deterministic solutions (fixed random seeds, fixed environments when needed) and verify answers using standard libraries (Pandas, NumPy, scikit-learn, statsmodels, etc.).
• Create prompts, model inputs, and reference responses for text-generation and SFT use cases (prompt + response writing, question answering, summarization).
• Write evaluation rubrics and rating guidelines for human reviewers to assess model outputs.
• Ensure tasks are computationally intensive enough that they cannot be solved manually in reasonable time.
• Include realistic business contexts (fraud detection, forecasting, optimization, customer analytics, risk) and sensible data constraints/performance considerations.

Requirements

You must demonstrate strong, hands-on expertise in applied data science, reproducible coding, and evaluation design. All requirements below are mandatory for screening.

• 5+ years of hands-on data science experience with measurable business impact.
• Expert Python for data science: Pandas, NumPy, SciPy, scikit-learn, statsmodels.
• Expert SQL: complex joins, aggregations, window functions, and database operations.
• Comfortable using visualization libraries for EDA and communication (Matplotlib; Seaborn a plus).
• Deep statistical and ML knowledge: feature engineering, model selection, evaluation, and error analysis.
• Proven ability to design deterministic, reproducible problems (fixed seeds, no stochastic ambiguity).
• Familiarity with big-data and scalable processing concepts (partitioning, memory and performance constraints).
• Experience with GenAI technologies (LLMs, retrieval-augmented generation, prompt engineering, vector DBs).
• Understanding of MLOps and model deployment workflows (packaging, reproducibility, monitoring basics).
• Experience with modern ML frameworks (TensorFlow or PyTorch); familiarity with LangChain is a bonus.
• Written English proficiency at C1+ level (able to write clear business problem statements and documentation).
• Availability to contribute roughly 10–20 hours per week during active project phases.

Who should apply

Apply if you’re a senior or principal data scientist, ML engineer, or applied researcher who enjoys designing realistic, reproducible problems and clear evaluation criteria. This role suits people who can translate business questions into computational tasks and provide robust, verifiable solutions.

• Ideal for practitioners who have led production analytics or ML initiatives and documented measurable outcomes.
• Good fit for contributors who want flexible, remote, part-time contract work and to influence how AI models learn from realistic data-science problems.

How it works — workflow and deliverables

Selected contributors will receive project briefs and templates. Typical deliverables include problem descriptions, synthetic or real-like datasets, runnable Python notebooks or scripts, solution notebooks with fixed seeds, testcases/verification code, and evaluation rubrics.

Work is reviewed for reproducibility, business realism, computational intensity, and clarity. Payment is hourly on a contract basis; individual projects will state scope, schedule, and acceptance criteria.

• Task types include prompt/response pairs, text generation examples, question-answering items, code-based problems, and summarization tasks.
• Deliverables must include reproducible code, clear instructions for graders, and verification artifacts demonstrating correct answers.

To apply

Create an OpenTrain account (free) and submit your profile highlighting relevant projects, sample notebooks, reproducible solutions, and a short note describing a representative data science problem you would author. Be explicit about your hourly expectations within the posted pay range and your typical weekly availability.

We will screen applicants for the technical requirements listed above and invite qualified candidates to a short technical review and sample task.