Nathaniel M.

Key Skills

Software

Appen

Other

Telus

CVAT

Remotasks

Top Subject Matter

Multimodal AI training datasets (emotion, temporal sequencing, visual cues)

Video emotion recognition

narrative/temporal sequencing

Top Data Types

Video

Image

Document

Top Task Types

Emotion Recognition

Action Recognition

Fine-tuning

Polygon

Text Generation

Freelancer Overview

Senior Image/Video Annotator – Multimodal AI Training (Appen, remote, US). Brings 9+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Appen, Other, and Telus. Education includes Doctor of Philosophy, University of Texas at Austin (2025) and Master of Science, University of Washington (2019). AI-training focus includes data types such as Video, Image, and Document and labeling workflows including Emotion Recognition, Action Recognition, and Evaluation.

Labeling Experience

Senior Image/Video Annotator – Multimodal AI Training (Appen, remote, US)

AppenVideoEmotion Recognition

Senior Image/Video Annotator for multimodal AI training specializing in emotion and visual cue labeling in short video clips. Annotated facial/body emotion categories and mapped ambient mood, lighting, and spatial relationships to support downstream multimodal understanding. Ensured consistency with an 80+ page annotation stylebook and ongoing calibration and QA processes. • Labeled emotions (e.g., happy, angry, confused, neutral) and intensity/visual cues from video frames. • Annotated mood/lighting conditions (bright, dim, backlit; suspenseful/cheerful/somber) and spatial relations (behind/beside/above/approaching). • Produced frame-accurate descriptions of scene transitions and character interactions. • Achieved 96% consistency with gold labels and maintained a 98% QA score for 8 consecutive months.

2022 - Present

Graduate Research Assistant – Computer Vision Annotation Lab (UT Austin, Dept. of Computer Science)

CVATVideoFine-tuning

Graduate Research Assistant supporting data labeling operations by designing annotation pipelines and training annotators for computer vision projects. Worked across multiple video understanding domains, creating structured labeling processes and improving annotator onboarding efficiency. Authored and maintained lab documentation used as best-practice guidance for video annotation. • Designed and maintained annotation pipelines for autonomous driving, assistive robotics, and video summarization. • Trained 12 undergraduate research assistants on CVAT and custom annotation tools. • Reduced onboarding time from 3 weeks to 5 days through improved training and workflow materials. • Authored a 35-page "Best Practices for Video Annotation" guide adopted by the lab.

2020 - 2025

Video Annotation Specialist – Temporal & Emotional Labeling (Lionbridge AI, remote, US)

OtherVideoAction Recognition

Video Annotation Specialist performing temporal and emotional labeling to train multimodal video AI systems. Labeled facial expressions and body-language cues frame-by-frame using an 8-category emotion set with intensity levels. Annotated narrative progression (exposition through resolution) and tracked up to 15 objects per clip while validating AI-generated outputs against human ground truth. • Emotion recognition: labeled joy, sadness, anger, fear, surprise, disgust, contempt, and neutral plus low/medium/high intensity. • Temporal storytelling: identified narrative beats for 30–90 second clips (exposition, rising action, climax, falling action, resolution). • Tracking: used bounding boxes and polygon masks to track up to 15 objects per clip across occlusions and re-entries with consistent IDs. • Validation/quality impact: compared model outputs to ground truth using predefined correctness outcomes and helped improve emotion recognition accuracy by 20%.

2020 - 2022

AI Output Validator – Image/Video Understanding (TELUS International, remote, US)

TelusImage

AI Output Validator for image/video understanding tasks validating AI-generated captions and ensuring they match the underlying visual content. Reviewed thousands of AI-generated outputs for factual correctness, grammatical structure, and semantic relevance. Conducted pairwise comparisons between model outputs and aligned with evolving annotation standards through feedback loops. • Reviewed 5,000+ AI-generated captions for images and videos, assessing correctness and relevance. • Evaluated side-by-side comparisons of two AI models using fixed outcome selections (Model A better, Model B better, both acceptable, neither acceptable). • Proposed guideline updates based on edge cases discovered during validation. • Audited junior annotator work by reviewing 500+ annotations per week and providing corrective feedback for recurring errors.

2018 - 2019

Freelance Image/Video Annotator (Upwork / Scale AI / Remotasks)

RemotasksImagePolygon

Freelance Image/Video Annotator completing high-volume microtasks for dataset labeling. Produced polygon-based annotations for autonomous vehicle training and facial landmark labels for emotion recognition. Also performed content moderation tagging and generated alt-text to support accessibility-focused AI models. • Drew polygons around vehicles and pedestrians for autonomous vehicle training datasets. • Labeled facial landmarks (eyes, nose, mouth corners) to support emotion recognition. • Identified and tagged unsafe or NSFW content in image datasets for filtering. • Wrote alt-text descriptions for images to train accessibility AI models.

2017 - 2018

Show more

Education

Work History