Diego D'aleman

Key Skills

Software

CVAT

Label Studio

OpenCV AI Kit (OAK)

Roboflow

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Image

Text

Top Task Types

Bounding Box

Classification

Computer Programming Coding

Data Collection

Fine Tuning

Freelancer Overview

AI Strategy Consultant with 14+ years bridging domain expertise and machine learning implementation. I specialize in designing data annotation frameworks, developing quality assessment protocols, and evaluating AI model outputs across Healthcare, Financial Services, and Computer Vision domains. My hands-on experience includes: designing bounding box annotation pipelines for object detection systems using CVAT and Roboflow; developing LLM evaluation rubrics and prompt engineering workflows for generative AI applications; creating labeled datasets for customer segmentation and predictive models in pharmaceutical marketing; and building algorithmic trading systems requiring extensive data labeling for backtesting validation. I bring deep subject matter expertise in healthcare/pharma (regulatory documentation, clinical terminology, HCP communications), quantitative finance (market data annotation, trading signal classification), and computer vision (object detection, image classification). Trilingual (English, Spanish, Italian) with advanced business education from Tuck and Wharton. Proficient in Python, C#, Label Studio, CVAT, OpenCV, and Roboflow.

Labeling Experience

Quantitative Trading Signal Classification

Internal Proprietary ToolingComputer Code ProgrammingClassificationAction Recognition

Built data annotation framework for trading signal classification and backtesting validation. Labeled market patterns including Keltner Channel breakouts, volume profiles, and price action signals across multiple timeframes. Developed quality control protocols for time-series data ensuring temporal consistency. Created ground truth datasets for strategy validation covering 3+ years of historical data. Implemented annotation pipelines in Python and C# for integration with cTrader and NinjaTrader platforms.

Built data annotation framework for trading signal classification and backtesting validation. Labeled market patterns including Keltner Channel breakouts, volume profiles, and price action signals across multiple timeframes. Developed quality control protocols for time-series data ensuring temporal consistency. Created ground truth datasets for strategy validation covering 3+ years of historical data. Implemented annotation pipelines in Python and C# for integration with cTrader and NinjaTrader platforms.

2024

Computer Vision - Chess Piece Detection

Opencv AI Kit OakImageBounding BoxPolygon

Designed end-to-end object detection pipeline for real-time chess piece recognition. Created annotation guidelines for 12 piece classes across varying board positions and lighting conditions. Developed bounding box labeling workflow with quality control checkpoints, achieving 95%+ annotation consistency. Specified model requirements for mobile deployment including latency constraints and accuracy thresholds. Dataset included 2,000+ annotated images with multi-angle captures and occlusion scenarios.

Designed end-to-end object detection pipeline for real-time chess piece recognition. Created annotation guidelines for 12 piece classes across varying board positions and lighting conditions. Developed bounding box labeling workflow with quality control checkpoints, achieving 95%+ annotation consistency. Specified model requirements for mobile deployment including latency constraints and accuracy thresholds. Dataset included 2,000+ annotated images with multi-angle captures and occlusion scenarios.

2024

LLM Evaluation & Prompt Engineering Framework

Label StudioTextClassificationText Generation

Developed comprehensive evaluation framework for generative AI outputs across multiple dimensions: factual accuracy, response coherence, instruction adherence, and safety compliance. Created rubrics and annotation guidelines for human evaluators assessing LLM responses. Built prompt optimization workflows iterating through systematic A/B testing. Labeled 500+ prompt-response pairs for quality benchmarking. Established inter-annotator agreement protocols achieving 90%+ consistency scores.

Developed comprehensive evaluation framework for generative AI outputs across multiple dimensions: factual accuracy, response coherence, instruction adherence, and safety compliance. Created rubrics and annotation guidelines for human evaluators assessing LLM responses. Built prompt optimization workflows iterating through systematic A/B testing. Labeled 500+ prompt-response pairs for quality benchmarking. Established inter-annotator agreement protocols achieving 90%+ consistency scores.

2023 - 2024

Healthcare Customer Segmentation & NLP Analytics

Label StudioTextClassificationEmotion Recognition

Created labeled datasets for HCP (Healthcare Professional) segmentation models in pharmaceutical marketing. Developed taxonomy of 15+ customer segments based on engagement patterns, specialty, and prescribing behavior. Annotated customer interaction data including email responses, call notes, and digital engagement signals. Built classification guidelines for sentiment analysis of HCP feedback. Dataset included 10,000+ labeled interactions enabling ML-powered personalization achieving 35% improvement in engagement metrics.

Created labeled datasets for HCP (Healthcare Professional) segmentation models in pharmaceutical marketing. Developed taxonomy of 15+ customer segments based on engagement patterns, specialty, and prescribing behavior. Annotated customer interaction data including email responses, call notes, and digital engagement signals. Built classification guidelines for sentiment analysis of HCP feedback. Dataset included 10,000+ labeled interactions enabling ML-powered personalization achieving 35% improvement in engagement metrics.

2022 - 2023

Education

Work History