Gulf Arabic Speech Transcriber

Transcribed Gulf Arabic and MSA audio in full verbatim, accurately labeling speakers, timestamps, and overlapping dialogue

Created high-quality, diverse training data across multiple annotation verticals to power the development of advanced computer vision, OCR, and data-extraction models.

• Polygon Image Segmentation (Cropping Spaces): Applied complex polygon segmentation techniques to 46 document images, isolating pages from cluttered backgrounds and correcting for image skew and rotation to create clean, machine-readable inputs. 4 mins/image.

• Bounding-box annotation (Box 20 Words): Annotated around 1,000 words across 50 tasks [English, Arabic, Hebrew, Hindi, CJK (Chinese, Japanese, or Korean)]; 7–8 min/task, 2–3 words/min. Produced tight, rotated boxes for angled and handwritten text.

• Complex document labeling (Document Derby): Annotated 5 diverse, multilingual documents (English/Japanese), dedicating an average of 48 minutes per document to apply a complex schema of up to 50+ detailed labels to layouts like invoices and medical forms.

• Text Legibility Classification (Can You Really Read Me?): Executed high-speed OCR data validation across Arabic, English, CJK (Chinese, Japanese, or Korean), and Indian scripts — classified 268 text samples at an average speed of 8 tasks per minute. Applied precise labels based on clarity (blurry), completeness (occluded), and stylization (unknown script).

• Video Content Analysis & Timestamping (Commercial Labeling v3): Performed detailed analysis of 32 video segments at an efficient pace of 1.5 minutes per task. Applied precise, frame-level intro/outro timestamps and performed multi-level content classification (e.g.,

• Commercial, TV Promo). Required nuanced decision-making based on a complex rubric, including hierarchical brand/product identification and context-based rules for streaming service promotions.

• Image Quality Analyst (Blurry or Not): Validated the quality of 96 multilingual documents (Arabic, English, Hindi, Hebrew, Japanese), executing a binary classification task at an efficient pace of 4.2 tasks per minute to reject blurred or corrupted images and ensure a high-quality input for machine learning models.

• Annotation Quality Assurance & Review (Getting Your Docs in a Row): Served as a key quality reviewer, auditing 30 pre-annotated documents at an efficient pace of 2 minutes per task. Applied expert knowledge of a complex annotation schema (20+ labels, including Tables, KV Areas, and Figures) to make nuanced "Approve/Reject" judgments on multilingual submissions, achieving a 100% task acceptance rate for quality assurance work.

• Process-Driven Quality Assurance: Developed and implemented comprehensive checklists based on detailed project instructions to ensure methodical and consistent application of annotation rules, resulting in a 96.6% overall task acceptance rate across 527 submissions.

Skills: Multimodal Data Annotation (Image, Text, Video) · Multilingual Labeling [English, Arabic, Hebrew, Hindi, Chinese, Japanese, Korean (CJK)] · Quality Assurance (QA) & Review · Bounding-Box · Polygon Segmentation · Complex Document Labeling · Video Timestamping · Attention to Detail · Image Quality Assessment

Project Goal: To master the professional standards of audio transcription and QA by applying industry-leading guidelines to a variety of Arabic audio samples.

• Executed clean verbatim transcription of a 3-minute, Saudi Arabic interview, meticulously adhering to GoTranscript's style guide to achieve a self-assessed 98% accuracy score.

• Performed a critical QA role by auditing and correcting 10 AI-generated (HappyScribe) transcripts in both MSA and Lebanese Arabic, reformatting them to meet the rigorous GoTranscript quality standards for STT model training.

Skills: Audio Transcription (Arabic) · Transcription QA · Dialect Comprehension (Saudi, MSA, Lebanese) · Guideline Adherence · Attention to Detail · Timestamping · Self-Assessment & Quality Control

3-Min Saudi Arabic Transcript: [link]

10 Arabic QA Transcript Samples: [link]

• Reviewed webpages and labeled quality and status according to a complex multi-point guideline to train AI search ranking models
• Achieved a throughput of 125 pages per hour (88 pages in 42 minutes) while adhering to client policies and QA standards

Core Achievement: Built an e-commerce business from the ground up on the Amazon KDP platform, managing the complete product lifecycle from market research and design to publishing and performance analysis.

Key Accomplishments & Skills:

• Market & Keyword Research: Utilized BookBolt to conduct in-depth market analysis, identifying high-demand, low-competition niches for puzzle books. Performed extensive keyword research to target specific customer segments, including adults, seniors, and kids.

• AI Model Analysis (R&D): Conducted a comparative analysis of multiple Large Language Models (LLMs) to optimize the word list generation process. Performed a quantitative evaluation of LLM performance based on data accuracy, relevancy, and efficiency to select the optimal tool.

• Product Design & Development: Conceptualized and developed multiple word search puzzle books. Managed the entire design process, creating professional, print-ready covers and interior layouts using specialized software (BookBolt) to meet Amazon's strict technical specifications.

• E-commerce Merchandising & SEO: Wrote compelling, SEO-optimized titles, subtitles, and descriptions to maximize visibility and ranking within Amazon's A10 search algorithm. Differentiated products by creating various formats, including hardcover, paperback, and large print.

• Performance Tracking & Analytics: Developed and maintained a comprehensive analytics spreadsheet to track key performance indicators (KPIs) and derive actionable insights to guide future product strategy.

Skills: Market Research (BookBolt) · Keyword Analysis · AI Tool Evaluation (LLMs) · Product Design · E-commerce SEO · Data Tracking & Analytics · Project Management

Portfolio: [link]

Engineered 20+ advanced AI prompts to optimize LLM outputs across diverse domains, including SEO content, high-converting book covers and descriptions, fiction/nonfiction book summaries, customized professional CVs, and large word search lists (500 words each)

• Developed a suite of prompts for high-performance content generation, focusing on SEO optimization for first-page article ranking and high-conversion copywriting for blog excerpts and e-commerce platforms (Amazon KDP).

• Engineered sophisticated prompts for complex data synthesis, successfully summarizing full-length fiction/non-fiction books and technical YouTube videos into concise, structured formats.

• Applied prompt engineering principles to career development, creating a systematic workflow to generate, iterate on, and quality-check world-class CVs, LinkedIn profiles, and other professional branding materials.

• Leveraged text-to-image prompting techniques to conceptualize and generate high-converting cover designs for Amazon KDP puzzle books, testing various artistic styles and compositions to align with target market aesthetics.

Skills: Prompt Engineering · LLM Evaluation & Optimization · Content Generation · Data Synthesis · Text-to-Image Prompting · Iterative Testing · Workflow Automation

Portfolio: [link]

Core Achievement: Built and grew two affiliate blog sites from scratch, engineering a comprehensive SEO project management system to move from initial concept to first-page Google rankings.

Key Accomplishments & Skills:

• Data-Driven Niche Validation: Developed a sophisticated framework in Sheets to forecast niche profitability, systematically evaluating opportunities against 15+ criteria (KD, Search Volume, EPMP, YMYL risk). Utilized this system to analyze macro-niches and strategically select high-potential verticals.

• SEO Performance Analysis & Optimization: Systematically tracked SERP rankings for all target keywords on a dedicated progress sheet. Analyzed this performance data to identify underperforming content, diagnose on-page issues, and execute iterative updates that successfully improved organic rankings.

• Full-Funnel Keyword Strategy: Executed an end-to-end keyword research process, identifying thousands of potential keywords and filtering them down to low-competition, high-intent topics with a clear path to achieving topical authority.

• Content Creation & SEO Copywriting: Wrote, edited, and published dozens of long-form articles, securing multiple first-page rankings by integrating best-practice SEO into engaging, user-focused content.

• A/B Testing & CRO: Designed and ran A/B tests on on-page elements, such as CTA text and banner placement, analyzing performance data to improve click-through rates and affiliate conversions.

Skills: SEO Strategy · Keyword Research (Semrush) · Content Strategy & Creation · SEO Copywriting · SERP Tracking · Data Analysis (Google Sheets, GSC) · A/B Testing & CRO · Affiliate Marketing · Project Management · WordPress

First Affiliate Blog Site (WayBack Machine): [link]

Second Affiliate Blog Site (WayBack Machine): [link]

View Niche Research & Brand Assets: [link]

• Executed five diagnostic tests (including turn ratio, winding resistance, power factor) on six high-voltage power transformers to assess operational health and compliance with specifications
• Operated the Megger TRAX Multifunctional Transformer & Substation Test System to record test data for final technical analysis and training report
• Adhered to strict field safety protocols in a high-voltage industrial setting