Generative AI Specialist - Flexible Hours
innodata • Remote - Nebraska
Education
Bachelor
Type
Hourly
Pay Rate
$15/hr
Listed
123d ago
About this role
From the innodata listing
Innodata (Nasdaq: INOD) is a global data engineering company. We believe that data and Artificial Intelligence (AI) are inextricably linked. Our mission is to enable the responsible advancement of artificial intelligence by providing the data, evaluation frameworks, and human expertise required to build AI systems that can be trusted at scale. We provide a range of transferable solutions, platforms, and services for Generative AI / AI builders and adopters. In every relationship, we honor our 36+ year legacy delivering the highest quality data and outstanding outcomes for our customers. Scope of the Role: At Innodata, we’re partnering with the world’s leading technology companies to build the future of generative AI and large language models (LLMs). We’re on the lookout for smart, savvy, and curious Generative AI Specialist to join our global contributor community as part of our Subject Matter Expert (SME) on Demand program. This is not a traditional full-time role. It’s a part-time, remote, flexible, project-specific opportunity designed for those who want to make a real impact—on their schedule. Whether you're a writer, linguist, educator, researcher, or just deeply passionate about language and logic, this role lets you contribute to cutting-edge AI development while maintaining control over your time. You’ll be helping LLMs learn the intricacies of language and reasoning—not just how to write, but how to think. If you’ve ever dreamed of shaping the intelligence behind tomorrow’s technology, this is your chance. This is more than just a gig—it’s a rare chance to help shape the future of AI from anywhere in the world, on your own terms. What You’ll Own: Rating/assessing the performance of AI models or algorithms based on their output or behavior through a set of evaluative questions. Labeling elements of a piece of content rather than the content as a whole. Assigning predefined categories or labels to items. Evaluating the perceived quality and/or appropriateness of content Generating labels to advance understanding of a concept, trend etc. Creation of additional training data for machine learning models by applying transformations to the original data, such as modifying images (rotation, flipping, cropping), generating new text (paraphrasing, summarization), or altering audio/video signals (speed modification, pitch shifting) to reduce overfitting and increase dataset diversity. Reviewing data and identifying whether or not a product feature works as intended based on the project's guidelines. Labeling model outputs to identify if a piece of content is or isn't something. Examples: identify clickbait; identifying gaming videos; identifying branded content. Ordering or ranking items based on a set of preferences or criteria. Creating prompts or questions that will be used to generate responses from a language model or other AI system. Projects that evaluate the relevance of content based on a relevancy scale (1-3, 1-5, etc.). Generating responses to prompts or questions using a language model or other AI system. Rewriting existing text while preserving the original meaning, often to improve clarity or style and adherence to guidelines. Producing concise summaries of longer pieces of text or data. Converting spoken language or audio content into written text. Converting text or spoken language from one language to another. Gathering and compiling various forms of data to be used for training, evaluating, or fine-tuning the AI models. This may include text, images, videos, audio files, or other types of digital content. You’ll Thrive in This Role If You Have: A Bachelor’s degree or higher in a humanities specialization is required. Advanced degrees are strongly preferred (Master’s or PhD) Professional or Expert level proficiency (C1/C2) in English The expected hourly salary range for this position is $15 p/hour, based on experience, skills, and qualifications. Please be aware of recruitment scams involving individuals or orga
Requirements
- Must be eligible to work in Remote - Nebraska
- Fluent proficiency in English (Written & Verbal)
- Reliable high-speed internet connection
Interview Prep
Sample questions for a Generative AI Specialist role, written in-house to help you prepare.
How do you decide between fine-tuning a pretrained model and training a smaller model from scratch for a generative task?
The decision comes down to data volume, latency budget, and how far the target task sits from the base model's original training distribution. Fine-tuning wins when labeled data is scarce and the task is close to what the base model already does. Training from scratch only pays off when you have enough domain data to justify the compute cost and need tight control over model behavior.
What causes mode collapse in generative models and how do you diagnose it?
Mode collapse shows up as a model producing a narrow range of outputs regardless of varied input prompts. It is diagnosed by sampling a large batch of outputs and measuring diversity metrics like embedding distance or n-gram overlap rather than eyeballing a handful of examples. Common causes include an overly aggressive learning rate, a discriminator or reward signal that is too easy to satisfy, or training data that is itself low-diversity.
How do you evaluate whether a generative model's outputs are factually grounded versus fluent but wrong?
Fluency and correctness are separate axes and need separate metrics. Grounding is checked by tracing claims back to source documents when retrieval is involved, or by running outputs through a fact-checking pass against a trusted reference set. Fluency alone, measured by perplexity or human read-through scores, will not catch confident-sounding errors, so both need to be tracked and reported separately.
What tradeoffs come with using reinforcement learning from human feedback versus supervised fine-tuning alone?
Supervised fine-tuning is cheaper and more stable but only teaches the model to imitate the examples it saw. RLHF lets the model optimize for a broader notion of quality captured in a reward model, which generally produces more helpful and better-calibrated outputs, but it adds training instability, reward hacking risk, and a much heavier annotation pipeline to build the reward model in the first place.
Why this role
At $15/hr, this Generative AI Specialist - Flexible Hours position pays for what you already know about Data Science. The AI training side of the job is covered during onboarding.
Talent pool
We're light on Data Science candidates
We've matched 9 people with a Data Science background against 293 Data Science listings we've tracked, so most go out without one. Set up a profile and we'll consider you for a role like this one.
Set up your profileSkills and categories
Explore other opportunities in related specializations:
Related jobs
Browse All Jobs from innodata
Discover more opportunities on innodata that match your skills and interests.
View All innodata Jobs →Verified Reviews
Community Reviews
Share your experience with innodata
Help other candidates make better decisions by leaving a review.
Sign in to leave a reviewLeave your review
Common questions
What does task-based AI training work look like?
Practical, hands-on data work: recording short videos, categorizing images, rating text responses, or analyzing data. Tasks are designed to be short and distinct, typically 5 to 60 minutes each.
What does asynchronous AI training work mean in practice?
No set hours, no check-ins, no meetings. You log in when you want, pick up an available task, complete it, and submit; nobody is waiting on you in real time. That's different from remote employment, where you're expected online during business hours. The tradeoff: you're competing with others for available tasks, so an empty queue means there's simply nothing to do until more work is released.
What does Data Science work look like for a Generative AI Specialist - Flexible Hours?
Tasks here are scoped to Data Science, not generic labeling. As a Generative AI Specialist - Flexible Hours, expect to draw on real domain judgment (evaluating outputs, correcting errors, or providing expert reasoning specific to Data Science) rather than following a one-size-fits-all rubric. If you don't have hands-on Data Science background, this is likely not the right listing to start with.
What specific skills does this listing call for?
English and Expert are named directly in the listing. If you don't have hands-on experience with these, expect the screening process to test for them directly rather than accepting adjacent experience as a substitute.
How much does this specific role pay?
This listing is posted at $15/hr, an hourly rate. Pay can change between when we last checked the listing and when you apply, so confirm the current number on the platform's own application page before committing time.
Can I apply from outside Remote - Nebraska?
This specific role is open only to people based in Remote - Nebraska. If you are somewhere else, applying is unlikely to lead to an offer even if you pass the assessment, because the restriction is usually about where the work can legally be contracted rather than your skills. Read the full description for any tax-residency or right-to-work caveats before you apply.