DataAnnotation: qualification decision and original diagnostic design
_ferdo Paths — planning specification v1 • 2026-11-15_
Stage: P0 and P2 design complete; no practice questions authored, no diagnostic run, no external application or actual Starter Assessment started. This is NOT an official or replicated DataAnnotation test.
1. Scope and qualification-track distinction
DataAnnotation's official FAQ names General, Multilingual, Coding, STEM, Professional as qualification tracks. Its role directory additionally groups jobs under Data & ML, Engineering, Economics, and other fields. The role categories do not establish that a separate Starter Assessment exists for each. Verify actual choices and eligibility on the signup screen, without consuming the one-attempt assessment.
Starter Assessment: one attempt, no retakes per FAQ. No applicant-paid signup fees. Further qualifiers may become available after acceptance; retake and compensation rules are test-specific. No acceptance or earnings guarantee.
2. First decision framework — non-personal template
| Option | Assessment status | Diagnostic evidence to establish |
|---|---|---|
| General | Published main track | nuanced reasoning, source-checking, clear English comparison |
| Coding | Published main track | independent algorithms, debugging, tests, complexity analysis |
| Multilingual | Published main track | faithful translation, cultural context, precise bilingual explanations |
| Data Scientist | Advertised job family; separate initial track unconfirmed | statistical critique, causal reasoning, Python/SQL/notebooks |
| STEM/Professional | Published main tracks, subject-specific credential rules | independent subject credentials, relevant expertise |
Decision sequence: (1) Confirm actual eligibility and options; (2) independently measure performance; (3) consider career fit and risk; (4) explicitly approve a single first attempt. Salary advertising is not a reliable first-attempt choice criterion.
3. Original diagnostic blueprint (questions not yet written)
| Unit | Planned time | Skills / item formats | Output |
|---|---|---|---|
| D0 Instructions, policy, baseline | 15–20 min | instruction comprehension, confidentiality, tool restrictions | policy checklist |
| D1 General evaluation | 45–60 min | compare model responses, identify false claims, assess sources, explain corrections | written evidence-based feedback |
| D2 Coding | 60–75 min | Python debugging, JavaScript/algorithm reasoning, edge-case tests, complexity | independently verified code judgments |
| D3 Data Science | 45–60 min | flawed analyses, sampling, confounders, leakage, SQL/pandas interpretation | methodological critique |
| D4 Turkish–English | 35–45 min | semantic equivalence, localization/register, cultural nuance | bilingual assessment |
| D5 Synthesis | 20–30 min | compare strengths/gaps, analyze critical errors, select track | private readiness report |
Total design estimate across all units: 3h40–4h50, split across days. This is a self-study plan, not the provider's actual test duration.
4. Educational rubric (not DataAnnotation's scoring)
- Correctness and substantive error detection: 35/100
- Reasoning, methodology, and defensible explanations: 25/100
- Validation, tests, edge cases and evidence: 20/100
- Instruction and constraint compliance: 10/100
- Clarity and precise written English: 10/100
Critical errors are flagged separately (fabricated facts, unsupported conclusions, invalid tests, wrong causal inference, instruction violations). Optional internal study thresholds may be considered later; never present these as provider acceptance thresholds.
All prompts and solutions will be original. Initial diagnostic work is completed independently, without AI assistance, to measure individual skills. Follow the real site's tool-use rules for any eventual real assessment. Do not reproduce leaked, confidential or proprietary test questions.
5. Future authoring requirements — not started
- ☐ Create objective, original question briefs and private answer keys for each D-unit.
- ☐ Independently validate code, statistics and marking rubrics.
- ☐ Prepare distinct feedback for correctness, reasoning, constraints and communication.
- ☐ Specify secure learner progress and local/online assessment execution, without publishing responses or identifiers.
- ☐ Verify provider eligibility, terms, available tracks and assessment-specific restrictions from the signup screen.
- ☐ Run diagnostics only after a new explicit decision to start.
- ☐ Make the real one-attempt application decision separately, after diagnostic review.
6. Data boundaries
Public material may contain curriculum specifications, generic examples and empty rubrics. Store personally identifiable profile data, individual scores, application details, identity documents, payment details, screenshots and real assessment content in private systems only. No actual Paths deployment is implied by this source file.
Private canonical record: Genel Merkez → Projects/Paths/DataAnnotation P0 P2 Karar ve Diagnostik Tasarimi.
Source references: Official FAQ, Coding overview, Software Engineer, Data Scientist, Turkish Specialist.