Anwar S.

Key Skills

Software

Don't disclose

CVAT

Data Annotation Tech

Top Subject Matter

Egyptian Arabic speech and dialect transcription for TTS fine-tuning

Egyptian government document annotation for detection and Arabic OCR

Top Data Types

Audio

Document

Image

Top Task Types

Transcription

Object Detection

Freelancer Overview

Junior Machine Learning Engineer — The Address Investments (dataset curation and transcription for fine-tuning). Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include OpenAI Whisper and Don't disclose. Education includes Bachelor of Science, London South Bank University (2025). AI-training focus includes data types such as Audio and Document and labeling workflows including Transcription and Object Detection.

Labeling Experience

Junior Machine Learning Engineer — The Address Investments (document annotation for YOLO and PARSeq)

Don't discloseDocumentObject Detection

Annotated Egyptian government documents to create labeled training data for document understanding tasks. The work included labeling document regions to support object detection and applying text recognition on scanned inputs. These annotations were used to fine-tune models for detecting document elements and recognizing Arabic text.• Labeled National IDs, criminal records, and birth certificates for detection training.• Used YOLO-based labeling to support document detection objectives.• Prepared inputs for PARSeq to perform Arabic text recognition on scanned documents.• Built a labeling output set suitable for ML fine-tuning and evaluation workflows.

2025 - 2026

Junior Machine Learning Engineer — The Address Investments (dataset curation and transcription for fine-tuning)

AudioTranscription

Collected and curated Egyptian Arabic speech data to generate aligned training pairs for downstream fine-tuning. Performed audio transcription to convert spoken content into text aligned with the audio segments. Prepared curated datasets for dialect-specific model training and TTS fine-tuning.• Collected 8+ hours of Egyptian Arabic audio for dataset creation.• Used OpenAI Whisper to transcribe audio into text for aligned audio-text pairs.• Curated and structured the dataset to support XTTS fine-tuning on Egyptian dialect speech.• Supported dialect-focused audio-text preparation workflows for ML training pipelines.

2025 - 2026

Education

Work History