Nashid Shabazz

Key Skills

Software

Scale AI

Data Annotation Tech

Other

Appen

Toloka

Snorkel AI

Telus

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Text

Computer Code Programming

Audio

Video

Image

Top Label Types

Classification

Text Summarization

Computer Programming Coding

Prompt Response Writing SFT

Translation Localization

Evaluation Rating

Audio Recording

Text Generation

Emotion Recognition

Data Collection

Transcription

Object Detection

Diagnosis

Function Calling

Fine Tuning

Relationship

Question Answering

Tracking

Freelancer Overview

I am a Senior Software Engineer with extensive experience in AI data training and annotation, specializing in designing and optimizing workflows for labeling text, image, and voice samples across diverse machine learning pipelines. My background includes developing internal tools to automate annotation processes, engineering advanced prompts for software engineering and system design topics, and implementing robust quality review systems to ensure accuracy and technical correctness of labeled datasets. I am skilled in C#, .NET Core, Python, Angular, React, and cloud platforms like AWS and Azure, and have built dashboards to visualize annotation progress and dataset quality trends. My work frequently involves collaborating with AI engineering teams to define annotation standards, creating automated validation scripts to flag inconsistencies, and authoring documentation to support data workflows and model behavior analysis. I am passionate about improving data quality and efficiency, and have contributed to projects in enterprise software, NLP, and structured data domains.

Labeling Experience

AI Agent Evaluation Analyst

TolokaComputer Code ProgrammingDiagnosisComputer Programming Coding

I work on Mindrift/Toloka projects such as Tendem, evaluating AI agents and LLM tools on complex, multi‑step client tasks. This includes coding and debugging in a remote virtual environment, building and testing function‑calling / API workflows, running web searches and business data analysis, and comparing LLM outputs for quality and correctness. I create and refine prompts and responses, classify and rate agent behavior, perform online research and document analysis, and pick up partially completed tasks from other evaluators to finish or correct their work. I follow detailed task‑approach, rejection, and handoff guidelines to decide when to request more information, reject or hand off tasks, and document my methods and results clearly for clients.

I work on Mindrift/Toloka projects such as Tendem, evaluating AI agents and LLM tools on complex, multi‑step client tasks. This includes coding and debugging in a remote virtual environment, building and testing function‑calling / API workflows, running web searches and business data analysis, and comparing LLM outputs for quality and correctness. I create and refine prompts and responses, classify and rate agent behavior, perform online research and document analysis, and pick up partially completed tasks from other evaluators to finish or correct their work. I follow detailed task‑approach, rejection, and handoff guidelines to decide when to request more information, reject or hand off tasks, and document my methods and results clearly for clients.

2025 - 2025

AI Expert Contributor: DevOps

Snorkel AIComputer Code ProgrammingDiagnosisFine Tuning

I design and implement terminal‑bench style coding tasks for Snorkel AI projects such as Terminus v1 and v2. Work involves creating and analyzing tasks that run in Docker, WSL, Linux, and Windows environments using tools like git/GitHub, Python, Bash, PHP, Harbor, and custom CLIs. I write task specs, reference implementations, and automated checks, then submit tasks through Snorkel’s pipelines for AI and human review. I use OpenAI, Claude, and other LLM APIs to validate difficulty, correctness, and guideline compliance, and I iterate based on reviewer feedback to improve coverage of real‑world developer workflows.

I design and implement terminal‑bench style coding tasks for Snorkel AI projects such as Terminus v1 and v2. Work involves creating and analyzing tasks that run in Docker, WSL, Linux, and Windows environments using tools like git/GitHub, Python, Bash, PHP, Harbor, and custom CLIs. I write task specs, reference implementations, and automated checks, then submit tasks through Snorkel’s pipelines for AI and human review. I use OpenAI, Claude, and other LLM APIs to validate difficulty, correctness, and guideline compliance, and I iterate based on reviewer feedback to improve coverage of real‑world developer workflows.

2025

Online Task Contributor

TelusImageRelationshipClassification

I work on Telus International projects involving multi‑modal data: image, audio, and video analysis and classification, social media content review, and online/business search tasks. My work includes image classification and relationship tagging, video content classification and tagging, audio analysis and categorization, and answering questions about media or search results by following detailed guidelines. I also perform web lookups and business searches, collect and verify social media or web data, and create or evaluate short prompts and responses for LLM‑powered products, always applying project‑specific quality and safety standards. I also perform LLM‑assisted data retrieval by using tools like ChatGPT and Perplexity AI to research topics, aggregate results, and provide structured analysis data in line with project guidelines.

I work on Telus International projects involving multi‑modal data: image, audio, and video analysis and classification, social media content review, and online/business search tasks. My work includes image classification and relationship tagging, video content classification and tagging, audio analysis and categorization, and answering questions about media or search results by following detailed guidelines. I also perform web lookups and business searches, collect and verify social media or web data, and create or evaluate short prompts and responses for LLM‑powered products, always applying project‑specific quality and safety standards. I also perform LLM‑assisted data retrieval by using tools like ChatGPT and Perplexity AI to research topics, aggregate results, and provide structured analysis data in line with project guidelines.

2024

Data Labeling Specialist

AppenVideoClassificationText Generation

I work on multiple Appen/CrowdGen projects (including Adam, Denovo, JigglyPuff, and Apple Baseline – Music Search and Classification). My tasks include text, image, audio, and video classification; object detection and tagging; business and local search lookups; and detailed audio analysis, transcription, and audio tagging. I create and evaluate prompts and responses for LLMs, review other contributors’ prompts/responses for quality, and apply complex rubrics to rate model outputs, search results, and media relevance. I follow extensive project‑specific guidelines for music search, web relevance, and multi‑modal content, ensuring consistent, high‑quality annotations at scale.

I work on multiple Appen/CrowdGen projects (including Adam, Denovo, JigglyPuff, and Apple Baseline – Music Search and Classification). My tasks include text, image, audio, and video classification; object detection and tagging; business and local search lookups; and detailed audio analysis, transcription, and audio tagging. I create and evaluate prompts and responses for LLMs, review other contributors’ prompts/responses for quality, and apply complex rubrics to rate model outputs, search results, and media relevance. I follow extensive project‑specific guidelines for music search, web relevance, and multi‑modal content, ensuring consistent, high‑quality annotations at scale.

2023

AI Evaluator

LabelboxAudioClassificationText Summarization

I worked on Alignerr projects such as Scatter – Data Visualization, Gravity, and others, focusing on multi‑step audio and prompt evaluation. Tasks included audio annotation, transcription, classification and detailed analysis, audio recording and generation, and extended label/rubric creation for complex sound scenarios. I created and evaluated prompts and responses for LLMs, rated and reviewed model and human outputs, and used custom rubrics for nuanced classification. For some tasks I sourced sounds from the internet and YouTube, integrated them into generated audio, and then labeled and reviewed the results in Labelbox. I also reviewed other raters’ work and provided feedback to maintain consistency and quality.

I worked on Alignerr projects such as Scatter – Data Visualization, Gravity, and others, focusing on multi‑step audio and prompt evaluation. Tasks included audio annotation, transcription, classification and detailed analysis, audio recording and generation, and extended label/rubric creation for complex sound scenarios. I created and evaluated prompts and responses for LLMs, rated and reviewed model and human outputs, and used custom rubrics for nuanced classification. For some tasks I sourced sounds from the internet and YouTube, integrated them into generated audio, and then labeled and reviewed the results in Labelbox. I also reviewed other raters’ work and provided feedback to maintain consistency and quality.

2025 - 2025

Show more

Education

Work History