Rauf Ibna Rafiq
LLM Labeler | Data Annotation & QA | AI Training Data
Hire this person
Send a job offer directly to this candidate
About
Working data annotator across four live task types: document and OCR field verification, computer-vision labeling, frame-level video captioning, and QA on AI-generated map output. Used to long client guideline documents, daily accuracy and TAT targets, and second-pass QC where the job is catching what a model or an earlier annotator got wrong. Alongside annotation, runs a small AI automation agency for e-commerce and US service-business clients, so written English client communication, following client-specific instructions, and handling confidential customer data are daily work.
Escalates genuine edge cases instead of guessing, and adapts quickly when guidelines change mid-project.
Experience
AI Trainer at RYU Data Center (2025-01 – 2026-08)
- Annotated and classified computer-vision data across bounding box, lane classification, and transportation projects, working strictly to client guidelines and project SOPs.
- Ran quality checks on AI-generated map centerlines over satellite imagery, correcting topology errors and fixing previously annotated output that failed review.
- Met daily productivity and accuracy targets, and caught recurring labeling errors before they spread across a batch.
- Coordinated with team leads over Discord and email to resolve ambiguous cases and stay aligned on guideline updates and process changes.
Data Annotation & QA at Contract Platform Work (2024 – Present)
- Receipt prelabel verification: review AI pre-labels on printed receipts field by field against a 14-key schema, approve or reject each field, type verbatim corrections, and write an explanation naming every key corrected. Typical calls include routing card payments out of cash fields, separating tax from discount, preserving printed number formatting exactly, and rejecting totals not actually printed on the receipt.
- Video description annotation: caption clips under a strict style guide covering verb form, per-hand attribution in every clause, banned vague verbs, object-naming consistency, and a five-second rule for idle segments.
- Completed multi-step onboarding and rubric quizzes before going live on paid batches, and keep personal reference docs so rulings stay consistent from session to session.
Co-founder at AutoXinity (Present)
- Build AI and workflow automation for e-commerce brands and US service businesses in HVAC, dental, and med spa.
- Built a cash-on-delivery order screening system for an e-commerce client, flagging high-risk orders before dispatch to cut return-to-origin losses.
- Handle client support directly in English: intake, troubleshooting, and explaining technical problems in plain language to non-technical business owners.
- Work daily with customer order, contact, and payment records under client confidentiality expectations.
IT & Accounting Operations Specialist
Earlier Experience, Data & Detail
- ran 12-hour operational shifts maintaining client databases and financial reporting with zero data loss.
Data & Operations Management at Talukdar Washing Plant
- verified high-volume data flows and daily production reporting for accuracy.
Industrial Trainee, Control Room at Ashuganj Power Station Co. Ltd.
- monitored real-time system parameters and maintained critical data logs.
Education
B.Sc. in Electrical & Electronic Engineering – Hajee Mohammad Danesh Science & Technology University (HSTU)
