Tobi. D.

Key Skills

Software

Argilla

Label Studio

CVAT

Doccano

Other

Roboflow

AWS SageMaker

Top Subject Matter

LLM alignment for financial services

agentic coding assistance

safety red teaming

Top Data Types

Text

Image

Top Task Types

RLHF

Fine-tuning

Prompt + Response Writing (SFT)

Segmentation

Freelancer Overview

Software Engineer / AI Trainer (AI evaluation, RLHF/RL preference training, red teaming, and rubric-based verification) . Brings 6+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Argilla, Label Studio, and CVAT. Education includes Bachelor of Engineering, Fundación Universitaria Luis Amigó (2022). AI-training focus includes data types such as Computer Code, Programming, and Text and labeling workflows including RLHF, Fine-tuning, and Prompt + Response Writing (SFT).

Labeling Experience

Software Engineer / AI Trainer (AI evaluation, RLHF/RL preference training, red teaming, and rubric-based verification) — Softtek, Remote

ArgillaRLHF

Led RLHF preference ranking and DPO pair generation for an LLM alignment project, producing 12K+ ranked response pairs to improve reward model accuracy. Performed code generation review and reasoning step verification using custom rubrics to audit chain-of-thought outputs across JavaScript, TypeScript, Python, and SQL. Developed adversarial prompts for red teaming to uncover safety failures across prompt injection, bias, and harmful-content categories. • RLHF preference ranking and DPO pair generation from model responses • Reasoning step verification and rubric-based auditing • Red teaming prompt design and safety failure identification • Function/tool-use evaluation validation against schema and grounding constraints

2025 - Present

Junior Experience Technology Engineer / Data Labeling Specialist — Publicis Global Delivery (PGD)

CVATTextFine-tuning

Delivered domain expert text, document, and image annotations for advertising and creative AI projects to support fine-tuning of brand-safe generative models. Executed multilingual AI training workflows (English, Spanish, Portuguese) validating cultural nuance, tone, and translation quality. Conducted A/B preference testing with rubric-driven scoring to evaluate marketing-copy LLMs and improve human preference win rate. • Text, document, and image annotation for fine-tuning datasets • Multilingual translation and nuance/tone validation • A/B preference testing with rubric-based evaluation • Adversarial prompt crafting for bias and harmful-content screening

2024 - 2025

Front-End and Web Application Developer (annotation UI + prompt evaluation harnesses) — iFactum - Highweb & Page Group Inc.

Label StudioTextPrompt + Response Writing (SFT)

Built annotation UIs integrated with Label Studio and doccano to capture text annotation and preference data for editor ranking workflows. Implemented prompt engineering harnesses to A/B test product description prompts and formalize reusable prompt templates across multiple retail clients. Produced evaluation and benchmarking rubrics for AI-written product pages to reduce copy rejection via iterative quality checks. • Text annotation and preference capture via annotation interfaces • A/B prompt testing and reusable prompt template creation • Rubric-driven evaluation/benchmarking of AI-generated product pages • Integration of labeling workflows into review queues for hallucination audits

2023 - 2023

Full Stack Developer (annotation pipeline engineering and labeling coordination) — Sioma

CVATSegmentation

Engineered image annotation and segmentation workflows for agricultural computer vision tasks using Roboflow and CVAT, supporting crop disease detection model training. Built point cloud annotation and satellite imagery annotation pipelines on AWS to process large-scale drone captures for precision agriculture analytics. Coordinated distributed annotation teams with calibration tests and QA reviews to raise bounding-box IoU agreement. • Image segmentation and crop disease annotation workflows • Point cloud and satellite imagery annotation pipelines • Calibration testing and QA reviews to improve annotation agreement • Management of labeling jobs, quality scoring, and labeler coordination

2021 - 2022

Education

Work History