Eyisto Aguilar Trejo

Argentina

Eyisto José Aguilar Trejo

Data Scientist · PhD in Applied Sciences

Data Scientist with a PhD in Applied Sciences and expertise in statistical modeling, quantitative analysis, and complex data systems. Experienced in developing computational models, performing mid-scale data analysis, and extracting actionable insights from high-dimensional datasets with knowledge in Python, SQL, Power BI, and Excel, with hands-on experience in time-series analysis, stochastic modeling, feature engineering, and data visualization. Passionate about leveraging data-driven methodologies to support business decision-making and predictive analytics.

Portfolio

Projects with measured results. Click any of them for the details: methodology, charts, results and demos.

ROC-AUC on test

With session totals0.988
Session start only0.865
+ user history0.878

Axis starts at 0.5 (chance)

Technical challenge · Data Scientist

Conversion propensity model

Predicts which web sessions end in a purchase (~393k Google Analytics sessions). The first version scored an AUC of 0.988 using the session's own totals; rebuilt with only pre-decision information: AUC 0.878 and ×7.8 lift.

PythonLightGBMTemporal validationFeature audit
View the full project
Regular plant$27marked as rareRare plant$64×2.4 · Etsy, 4″ pot
Professional work · Plants Without Borders

Collectible plant price prediction

A model that suggests price bands for collectible plants from listings across several marketplaces. R² 0.23 on test with the root cause identified, plus an LLM (Gemini) extraction pipeline.

Pythonscikit-learnRandom ForestFeature engineering
View the full project
Actual history7-day forecast
Technical challenge · Forecasting

SKU-level sales forecasting

7-day forecast of daily sales for 20 products. A genuinely separate test set with WAPE 0.730 and a ~26% improvement over Seasonal Naive in the 28-day evaluation.

PythonLightGBMSHAPTime series
View the full project
Competition · Kaggle

AlixPartners Data Challenge: box catalog

Redesign of a fictional company's box catalog (427 products, 5 plants): from 204 to 59 box types and a validated saving of $20.9M per year (9.96% lower cost).

OR-Tools CP-SATMILPLarge Neighborhood SearchOptimization
View the full project
Published product

Capital Party

An events guide for salsa, bachata and nightclubs. A React Native app on the App Store and Google Play, fed by a pipeline that extracts events from flyers with Gemini.

React NativeFirebaseGoogle Cloud RunGemini API
View the full project
Own project · Full-stack

InvierteEdu: interactive financial education

An app built with React, Recharts and FastAPI: 12 modules on RSI, MACD, moving averages, P/E, EPS, ROE and dividends, with a live price chart for any stock. The app's interface is in Spanish.

ReactTypeScriptViteTailwind
View the full project

Professional experience

Data Scientist – Quantitative Modeling & Analytics (Consultant)

Oct 2024 – Present

Own project / Self-Startup · Remote

  • Developed statistical and computational models in Python to analyze large-scale complex datasets (~10,000+ nodes).
  • Performed exploratory data analysis (EDA) and time-series analysis to detect patterns, correlations, and emergent behaviors.
  • Designed quantitative metrics to monitor system states and validate model performance.
  • Applied stochastic modeling and simulation techniques to evaluate system dynamics under different scenarios.
  • Communicated findings through data visualizations and structured analytical reports.

Tech stack: Python (Pandas, NumPy, SciPy), Statsmodels, SQL, Power BI, Excel, Google Cloud / AWS

Jr. Data Scientist

Aug 2024 – Jan 2026

Plants Without Borders · Remote

  • Automated the extraction of product sheets (name, price, pot size) with the Gemini API over parsed HTML, independent of each site's structure, with JSON output into SQL.
  • Performed exploratory data analysis (EDA) to identify trends, anomalies, and data quality issues.
  • Built a price prediction model: baseline (DummyRegressor), linear regression, then Random Forest and gradient boosting, with a log target and a 70/15/15 split evaluated with RMSE.
  • Cleaned, transformed, and prepared datasets for analytical modeling.

Tech stack: Python (Pandas, NumPy, scikit-learn), Gemini API, Beautiful Soup, Selenium, SQL (PostgreSQL), Excel, Power BI

Research Assistant | PhD Researcher

Jun 2020 – Dec 2023

National University of San Martín · Argentina

  • Conducted numerical simulations of neural network models using Python.
  • Analyzed avalanche size distributions under varying parameter configurations to identify emergent patterns.
  • Performed statistical analysis of simulation outputs to evaluate system behavior.
  • Author of 4 articles in Physical Review E (one as first author) and an arXiv preprint.

Tech stack: Python, numerical simulation, statistical analysis

Professor

Mar 2018 – Feb 2020

University of the Andes · Venezuela

  • Taught undergraduate physics and quantitative methods courses.
  • Developed analytical problem-solving frameworks for technical subjects.

Technical skills

Programming & data analysis

  • Python (Pandas, NumPy, SciPy)
  • SQL
  • FastAPI
  • Jupyter Notebook
  • Git

Statistics & machine learning

  • Scikit-learn
  • LightGBM
  • SHAP
  • Statsmodels
  • Statistical inference
  • Hypothesis testing
  • Time-series analysis (ARIMA, autocorrelation)
  • Backtesting and temporal validation
  • Regression and classification models

Modeling & simulation

  • Stochastic modeling
  • Monte Carlo simulation
  • Numerical methods
  • Dynamical systems

Data processing

  • Data cleaning and transformation
  • Feature engineering
  • High-dimensional data
  • ETL
  • LLM-based data extraction (Gemini API)

Visualization & BI

  • Power BI (dashboards and KPIs)
  • Advanced Excel
  • Matplotlib
  • Seaborn

Cloud & tools

  • Google Cloud Platform (Cloud Run)
  • Docker
  • Firebase
  • AWS (basic)

Education

PhD in Applied Sciences and Engineering

National University of San Martín

2021 – 2023

Master's in Fundamental Physics

University of the Andes

2017 – 2018

Bachelor's Degree in Physics

University of the Andes

2009 – 2017

Doctoral Research Fellowship – CONICET

Argentina

Certifications and publications

  • Finite-size correlation behavior near a critical point: A simple metric for monitoring the state of a neural network — Physical Review E 106, 2022 (first author)
  • Similar local neuronal dynamics may lead to different collective behavior — Physical Review E 104, 2021
  • Self-tuned criticality: Controlling a neuron near its bifurcation point via temporal correlations — Physical Review E 107, 2023
  • Scale-free correlations in the dynamics of a small (N ≈ 10 000) cortical network — Physical Review E 108, 2023
  • School on Applications of Nonlinear Systems to Socio-economic Complexity — ICTP South American Institute for Fundamental Research, 2022
  • Courses on model-based neural systems & data-driven modeling — XIX Regional Congress on Statistical Physics and Applications to Condensed Matter

Languages

Spanish · NativeEnglish · AdvancedPortuguese · Basic

Let's talk about your next data project

Available for data analysis, statistical modeling and machine learning projects.