Gil L.

Key Skills

Software

Other

Top Subject Matter

AI image generation quality assessment and evaluation

Video annotation support for AI evaluation workflows

Prompt engineering and AI training/evaluation for image generation

Top Data Types

Image

Video

Text

Top Task Types

Tracking

Prompt + Response Writing (SFT)

Freelancer Overview

AI Trainer and Evaluator — Handshake (Project Hedgehog). Brings 6 months of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal and Proprietary Tooling. Education includes Bachelor of Science, University of Nevada, Reno (2025). AI-training focus includes data types such as Image, Video, and Text and labeling workflows including Evaluation, Rating, and Tracking.

Labeling Experience

AI Trainer and Evaluator — Handshake (Project Hedgehog)

TextPrompt + Response Writing (SFT)

Crafted and refined desired prompts to guide AI image generation toward specified outcomes. Tuned prompt instructions based on evaluation results from image and video checks. Used continuous multi-model comparison to revise prompting and improve reliability of the generated outputs. • Prompt engineering for target outcomes • Prompt iteration based on quality evaluation • Multi-LLM comparative evaluation • Guidance to improve generation consistency

2025 - Present

AI Trainer and Evaluator — Handshake (Project Hedgehog)

VideoTracking

Used bounding boxes, anchor tracking, and keyframe correction to review video content for annotation readiness. Applied pixel-perfect masking to adjust and isolate objects, supporting extraction of visual patterns. Coordinated evaluation steps to ensure annotations aligned with expected object locations and visual fidelity. • Bounding box review for objects of interest • Anchor tracking across frames • Keyframe correction for temporal alignment • Pixel-perfect masking for object isolation

2025 - Present

AI Trainer and Evaluator — Handshake (Project Hedgehog)

Image

Evaluated AI-generated images for visual quality, realism, and end-user preference using rubric-based review. Assessed whether outputs met desired outcomes and recommended prompt refinements to steer future generations. Performed iterative challenges across model failures to identify weaknesses and improve labeling guidance. • Quality check of generated images • Preference and realism evaluation against criteria • Iterative feedback loop to enhance prompt quality • Failure analysis to improve evaluation outcomes

2025 - Present

Education

Work History