Dereck Mendez — Building intelligent systems from data to decisions.
I build data and machine learning solutions that turn real-world problems into measurable, reproducible outcomes — from analysis and experimentation to reliable systems.
Hi, I'm Dereck.
My path into data has been shaped by a combination of business experience, analytics and software development. I enjoy understanding the problem behind the data before thinking about the model.
Today, I'm focused on growing as a Data Scientist and Machine Learning Engineer, with a particular interest in the engineering practices that make ML systems reproducible, maintainable and useful beyond experimentation.
More about me→
End-to-end projects solving real problems.
A selection of projects where I combine data analysis, statistical reasoning, machine learning and engineering to move from a business question to a working solution.
An end-to-end compensation benchmark for data and AI professionals worldwide: contract-tested ETL, statistical rigor (hypothesis testing, bootstrap confidence intervals), purchasing-power adjustment, business SQL queries and an interactive Power BI dashboard.
An end-to-end churn prediction system spanning EDA to production: model selection, experiment tracking, drift monitoring, champion/challenger retraining, explainable serving, and a business-value calculation to prioritize retention.
A Medallion-architecture data pipeline (Bronze → Silver → Gold) for e-commerce data, built with Polars and DuckDB and orchestrated with Airflow. A work in progress focused on demonstrating end-to-end data engineering.
The stack, organised by discipline.
Data & Analytics
Wrangling, exploring and modelling data in code.
BI & Visualization
Turning analysis into something people read and use.
Data Engineering
Storage, transformation and orchestration.
Machine Learning & MLOps
Training, tracking, versioning and serving models.
scikit-learn
LightGBMEngineering Practices
Version control, testing, CI and containers.
From business experience to data and ML.
Full experience →Working at the intersection of analytics, software and data-driven solutions, connecting real business problems with technical solutions while specializing in Machine Learning and MLOps.
Combined people management with People Analytics and BI: automated reports with Python and built Power BI and Looker Studio dashboards with data from Guatemala, Costa Rica and El Salvador.
Trained new hires in the inventory area and began building my own indicators and dashboards in Excel, along with tools to standardize processes.
Coordinated the inventory control team and oversaw counts, discrepancies and adjustments, making data-driven decisions to keep inventory reliable.
My first steps in data analysis: cycle counts in Excel, spotting shortages and surpluses, and investigating the root causes behind each discrepancy.
Started on the operations side, in direct contact with customers, learning firsthand how a branch works: the foundation for everything that came next.
What I can actually do.
Data Analysis
Turn business questions into structured analyses, using data to understand what is happening and why.
Data Visualization
Communicate analytical results through clear dashboards and visualizations designed for decision-making.
Machine Learning
Build and evaluate predictive models with strong baselines, appropriate validation and metrics aligned with the problem.
Forecasting
Model temporal patterns and evaluate forecasts using validation strategies that reflect how predictions will be used in practice.
MLOps
Build reproducible ML workflows around experiment tracking, versioning, testing, packaging and automated retraining.
Data Engineering
Design reliable data pipelines — ingestion, layered transformations and orchestration — so analytics and models run on clean, reproducible data.
Let's build something meaningful together.
I'm open to opportunities where I can contribute across Data Science, Machine Learning, analytics and the engineering behind reliable data products.
Whether you're looking for someone to analyse a problem, build a model or help turn an experiment into a more reliable system, I'd be happy to connect.