Tauseef FayyazA to Z
Glossary
Every concept in this sheet, plus the smaller terms it uses. Select a concept to open its card.
#
- 2026 demand evidence
- Upwork In-Demand Skills 2026: AI data annotation and labelling demand up 154 percent year over year.
D
- Demonstrations (SFT data)
- A prompt paired with the ideal answer, written by a person.
E
- Effective hourly rate
- What you actually earned divided by all the hours you spent, including reading guidelines, unpaid tests and rework.
- Eval (evaluation)
- A set of test prompts plus a way to score the answers, by code or by people using a rubric, to measure how good a model is at something.
I
- Independent contractor
- You work for the platform as a contractor, not an employee: no paid leave, health cover or pension, and you handle your own tax.
L
- Labelled data
- Examples with the correct answer attached, such as a photo marked "bus" or a review marked "negative".
P
- Pay that varies
- Platforms set rates by expertise, project difficulty and location, and show the rate before you accept a project.
- Preference data and RLHF
- Records of which of two or more answers a person preferred.
- Project-based work
- Work opens when a client needs data and closes when they have enough.
R
- Red teaming (safety testing)
- Deliberately trying to make a model misbehave, such as giving harmful instructions or leaking data, under project rules, and reporting what worked.
- Reward model
- A model trained on human preference data to predict which answer people would prefer, giving each answer a score.
S
- Sensitive content work
- Projects reviewing harmful material, such as violence, hate or self-harm, so models learn to refuse or flag it.
T
- The AI data supply chain
- AI labs hire data vendors (Scale AI, Appen, TELUS Digital, Toloka, Labelbox, Turing), which recruit workers through platforms such as Outlier, CrowdGen, Mindrift and Alignerr.