Hire this AI Trainer
Sign in or create an account to invite AI Trainers to your job.
Mercor
Labelbox
Text
Audio
Document
Transcription
Red Teaming
RLHF
Evaluation/Rating
Computer Programming/CodingLinguistics Expert (Arabic) - Innodata. Brings 2+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include Don't disclose, Internal, and Proprietary Tooling. Education includes Bachelor of Science, Zewail City of Science and Technology (2028). AI-training focus includes data types such as Text and labeling workflows including Evaluation, Rating, and Transcription.
TextEvaluation/RatingEvaluated medical AI responses to detect misinformation, hallucinations, and factual inaccuracies. Leveraged my academic background in Biomedical Sciences to perform rigorous qualitative analysis and fact-checking of healthcare-related prompts. Tasks involved comparative analysis, strict prompt alignment adherence, and rating model outputs based on medical accuracy and safety standards.
TextRLHFConducted blinded A/B evaluation of model outputs and classified failure modes using a strict loss taxonomy including Knowledge, Reasoning, Depth, and style. Wrote concrete preference explanations contrasting winning and losing responses, explicitly describing the impact of reasoning or formatting errors on the user’s query. Performed targeted re-annotation passes to expand shorthand feedback into fully articulated judgments while maintaining human-only drafting protocols. • Blinded A/B evaluation and loss-taxonomy failure classification • Preference reasoning authoring (2–4 sentence explanations) • Targeted re-annotation for clearer rationale quality • Human-only drafting (zero LLM assistance) and high-throughput consistency
TextRed TeamingAssessed AI-generated responses using the ICON-REAL framework, evaluating intent fulfillment, factual truthfulness, logical coherence, and relevance. Performed evidence-based fact-checking by independently researching claims and citing publicly accessible, non-paywalled sources. Authored precision Strengths and Areas of Improvement with exact response excerpts, and executed preference ranking to compare multiple model outputs for alignment. • RLHF-style quality and truthfulness assessments (ICON-REAL) • Fact-checking and hallucination identification with sources • Strengths/AOIs authored from exact excerpts and CoT/format deviations • Preference ranking and alignment justifications (Correct/Concise/Clear/Harmless/Lawful/Accessible)
TextTranscriptionReviewed and approved labeled Air Traffic Control audio rows under strict production standards. Validated transcriptions for accurate adherence to casing hierarchies, callsigns, acronyms, NATO phonetics, and facility names per project directives. Segmented speakers (Pilot vs. ATC vs. Unknown) and enforced rejection/skip protocols for indecipherable audio or poor label quality. • FAA phraseology and acoustic evaluation compliance checks • Transcription formatting validation (callsigns, acronyms, phonetics) • Speaker role segmentation and turn-taking boundary checks • Enforced skip/rejection logic for overlap and high-static audio
Text
Prompt + Response Writing (SFT)Reviewed LLM outputs for factuality, cultural appropriateness, safety, and linguistic correctness in Modern Standard Arabic with dialectal considerations. Performed linguistic annotation and multilingual corpus QA to support training and data validation. Authored iterative feedback and documented bias patterns and failure modes to strengthen model alignment and dataset quality. • Linguistic annotation and multilingual corpus QA • Accuracy, safety, and cultural appropriateness checks • Bias/failure-mode identification with recommendations • Corrected Arabic transcriptions and timing using SRT-based workflows Authored high-quality, complex Arabic prompts and model responses for Supervised Fine-Tuning (SFT). Ensured exceptional linguistic accuracy, cultural relevance, and contextual understanding in Arabic NLP. Responsibilities included generative AI content creation, prompt alignment, text summarization, and maintaining high standards for AI model training datasets.
Bachelor of Science, Biomedical Sciences (Computational Biology)
ATC Reviewer (Quality Assurance)
LLM Evaluator (Live Mobile Experiences)
