Sean B.

Key Skills

Software

Appen

Clickworker

CloudFactory

CVAT

CrowdSource

Data Annotation Tech

Google Cloud Vertex AI

Figure Eight

HiveMind

Humanatic

iMerit

Img Lab

Labelbox

Mercor

Micro1

Mindrift

OneForma

OpenCV AI Kit (OAK)

Remotasks

SuperAnnotate

Snorkel AI

Scale AI

Surge AI

Toloka

V7 Labs

Don't disclose

Internal/Proprietary Tooling

Roboflow

Supervisely

Top Subject Matter

Artificial Intelligence & Machine Learning – LLM Evaluation & Alignment

Software Engineering & Computer Science – Code Annotation & Prompt Engineering

Cybersecurity & AI Safety – Adversarial Red-Teaming & Failure Mode Detection

Top Data Types

Text

Computer Code Programming

Image

Top Task Types

Bounding Box

Polygon

Classification

Entity (NER) Classification

Point/Key Point

Object Detection

Text Generation

Question Answering

Text Summarization

RLHF

Fine-tuning

Red Teaming

Transcription

Evaluation/Rating

Computer Programming/Coding

Data Collection

Function Calling

Prompt + Response Writing (SFT)

Segmentation

Freelancer Overview

Sean Boggs AI Specialist & Software Engineer 530-417-3999 | [email protected] | Rancho Cordova, CA | github.com/DiggityDooo PROFESSIONAL OVERVIEW Results-driven AI Specialist and full-stack software engineer with hands-on experience in model evaluation, RLHF annotation, adversarial red-teaming, and instruction tuning across multiple AI platforms. Combines a cognitive psychology academic background with deep technical proficiency in Python, TypeScript, and cloud-native architectures to deliver precise, multi-dimensional AI feedback at scale. Proven ability to identify model failure modes, engineer targeted alignment strategies, and ship production software tools — from open-source OSINT CLIs to Claude API-powered web applications. WORK EXPERIENCE AI TRAINER & AI SPECIALIST | AfterQuery, RemoteMAY 2026 – PRESENT • Spearheaded generative AI model evaluation and instruction tuning, applying cognitive science principles to systematically debug model hallucinations and implement targeted alignment strategies. • Designed and executed adversarial prompts to stress-test model behavior across reasoning, coding, and instruction-following domains, delivering structured feedback that improved output quality and reduced failure rates. • Collaborated cross-functionally with research teams to develop alignment strategies, leveraging full-stack engineering expertise to provide technically rigorous, reproducible model assessments. AI MODEL EVALUATOR | Mercor AI, Remote (Contract)JAN 2026

Labeling Experience

Code Prompt Engineering & Instruction Tuning — AI Training Data Production

Computer Code ProgrammingComputer Programming/Coding

Engineered high-quality prompt-response pairs and instruction-tuning examples for code generation model training, applying full-stack software engineering expertise across Python, TypeScript, JavaScript, and Bash. Designed precise, unambiguous task prompts covering algorithm implementation, API usage, debugging, and system design — and produced corresponding gold-standard solutions with structured annotations covering correctness, efficiency, style adherence, and edge-case handling. Contributed to SFT and RLHF pipelines by generating reproducible adversarial code prompts designed to surface model failure modes in reasoning-heavy and multi-step implementation tasks.

2026 - Present

LLM Adversarial Red-Teaming — Hallucination Detection & Safety Evaluation

TextRed Teaming

Designed and executed systematic adversarial prompt suites to stress-test large language model behavior across reasoning, coding, factuality, and safety domains. Identified and documented latent edge-case failure modes including hallucinations, instruction-following breakdowns, jailbreak vulnerabilities, and unsafe output patterns. Delivered structured failure reports with reproducible test cases used by research teams to target alignment interventions. Maintained a rigorous red-team methodology grounded in cognitive behavioral analysis, systematically probing model responses for bias, sycophancy, and policy violations at production scale.

2026 - Present

RLHF Preference Labeling — LLM Response Ranking & Pairwise Comparison

TextRLHF

Executed production-scale RLHF annotation including pairwise response comparison, ranked preference labeling, and multi-criteria quality scoring across LLM outputs spanning reasoning, instruction-following, mathematics, and dialogue tasks. Applied fine-grained rubrics evaluating helpfulness, harmlessness, honesty, groundedness, and stylistic adherence. Generated high-quality gold-standard training signals that fed directly into model alignment and performance optimization pipelines. Maintained high annotation throughput with strict inter-rater agreement standards and minimal supervision across ambiguous, open-ended labeling tasks at production scale.

2026 - Present

Multimodal Image Quality Evaluation — AI Model Training

ImageEvaluation/Rating

Performed high-volume image quality evaluation and relevance scoring for multimodal AI model training pipelines. Tasks included visual accuracy assessment, bounding-region relevance scoring, image-response alignment rating, and multi-dimensional quality rubric scoring across helpfulness, visual groundedness, and safety compliance. Produced structured calibration datasets and curated golden-answer sets that directly improved inter-rater agreement scores and downstream model alignment quality. Maintained strict guideline adherence and high annotation throughput across diverse visual task types throughout a AI Fellowship program.

2025 - Present

Education

Work History