Franco GutierrezSoftware Developer
Versión en español: ESEN

Franco Sebastian Gutierrez · Software Developer · Chaco, Argentina · Remote

Web, backend and machine learning.

I build APIs, internal tools and web applications with Go and React. I analyze data, train and evaluate models with Python, and integrate models into services with FastAPI and Docker.

View projects Contact meDownload CV GitHub
Web & backend
Go · FastAPI · React · SQL Server
Machine learning
Scikit-learn · LightGBM · TensorFlow
Data & infrastructure
Pandas · Redis · Docker · Linux

01 Software

Internal tools and web applications.

Two Go systems that automate administrative work, inventory and records, plus a React frontend demo.

01Internal tools

Inventory System

From Excel to SQL Server, with web-based inventory management.

Internship · Government House of the Province of Chaco

Real interface · Sample data
Real screenshot of the Inventory System with sample data.
Challenge
The inventory was searched and managed by hand in Excel spreadsheets.
Result
Web-based management with SQL Server: search, filters, loans and movement history.
My role
Solo development
Status
In advanced development
Scope
Go backend with SQL Server for equipment, supplies, loans and movement history. Imports and exports Excel to migrate and back up data.
Stack
  • Go
  • SQL Server
  • excelize
  • Docker Compose
Private code · internal tool
More details

Internal Go application to manage equipment, supply stock, loans and movement history from a web interface, with a summary and reports.

Problem
Searching and managing the inventory by hand in Excel made it hard to access the information and manage the devices.
Solution
A Go REST API with a web interface: CRUD, search, filters, sorting and pagination. Data lives in SQL Server; Excel remains for import and export.
  • Separate layers: HTTP handlers, services with the business logic and interchangeable repositories (SQL Server or Excel) behind the same interface.
  • SQL migrations and command-line tools to set up the database and migrate the data from the spreadsheets.
  • Stock movements (in, out, adjustment and transfer), loans that expire at the end of the workday, summary and reports.
  • Docker Compose with a hardened container: read-only file system, no extra privileges and rotated logs.
  • Go tests for handlers, services and repositories, plus frontend tests.

02Administrative automation

Records Generator

Administrative documents, generated from the web.

Internship · Government House of the Province of Chaco

Real interface · Sample data
Real screenshot of the Records Generator with sample data.
Challenge
Each record requires filling in data, validating it and producing the document in the right format.
Result
A web form generates the record as DOCX and PDF, with an automatic file name.
My role
Solo development
Status
Completed
Scope
HTTP API and web interface that generate DOCX files and convert them to PDF with LibreOffice.
Stack
  • Go
  • LibreOffice
  • Docker Compose
  • Proxmox
Private code · internal tool
More details

Go system that generates dynamic administrative records as DOCX and automates their conversion to PDF with headless LibreOffice.

Problem
Preparing administrative records requires filling in information, validating the data and producing documents in the required formats.
Solution
A web interface connected to an HTTP API generates the documents, names the files automatically and converts them from DOCX to PDF.
  • Go HTTP API and web interface to enter data and generate records.
  • LibreOffice in headless mode to automate the DOCX to PDF conversion.
  • Docker and Docker Compose to run the service deployed on Proxmox.

03Frontend / Visual demo

Horizonte Global

Landing page and CRM for a travel agency.

Freelance project

Frontend demo · Mock data
Screenshot of the original Horizonte Global dashboard, a frontend demo with mock data.
Challenge
Present the agency and organize clients and immigration procedures in a single interface.
Result
Multilingual landing page and navigable CRM in React, as a visual demo with mock data.
Status
Frontend demo
Scope
Multilingual landing page and navigable CRM in React with mock data; no backend or database.
Stack
  • React
  • React Router
  • Tailwind CSS
  • Vite
View repository
More details

Frontend prototype for a travel and international mobility agency. It includes a multilingual landing page and a CRM with clients, cases, documents, calendar and team, using mock data.

Problem
Presenting the agency and visually exploring how to organize clients and immigration processes in a single interface.
Solution
A navigable landing page and dashboard in React. The scope is a visual demo: no backend, database or real API integration.
  • Components for the landing page and dashboard, with navigation through React Router.
  • Tailwind CSS, Framer Motion and Lucide for styles, transitions and icons.
  • Mock data inside the frontend itself. The repository has no backend or database.

02 ML & Data

Models, inference and data analysis.

Three AnyoneAI projects: an image classification service, a comparison of credit risk models and an ELT pipeline.

04Computer vision

Image Classifier

Classification API with a pretrained ResNet50. Streamlit lets users upload images and Redis coordinates the prediction processing.

  • Python
  • FastAPI
  • Streamlit
  • TensorFlow / Keras
  • Redis
  • Docker Compose
View repository
Project interface · Inference example with a toucan
Streamlit interface of the classifier with an image of a toucan and the prediction toucan.
Technical approach
  • The interface, the API and the inference worker run as separate services with Docker Compose.
  • The worker takes jobs from a Redis queue and stores the predicted class and its score under the job ID.
  • Images are resized to 224 × 224 pixels and preprocessed for ResNet50. The top-scoring ImageNet class is returned.
  • The scope is integrating a pretrained model with an API and an upload interface; the available categories are those of ImageNet.

05Predictive models

Home Credit: credit risk

Analysis of 307,511 loan applications and comparison of logistic regression, Random Forest and LightGBM to estimate the risk of payment difficulties.

0.7551ROC AUC · ValidationLightGBM · 49,202 applications in validation.

  • Logistic Regression
  • Random Forest
  • LightGBM
View repository View evaluation
Training dataset exploration · 8.03% with payment difficulties
Notebook chart: distribution of the 246,008 training applications, with 8.03% of cases with payment difficulties.
Technical approach
  • Dataset: 246,008 applications for training and 61,503 for test. 20% of the training set is held out for validation: 196,806 to train and 49,202 to evaluate, with seed 42.
  • Validation ROC AUC: 0.6772 with the initial logistic regression, 0.7465 inside a pipeline, 0.7323 with a tuned Random Forest and 0.7551 with LightGBM.
  • LightGBM uses early stopping and reaches its best iteration at round 218. The result corresponds to the validation set used during tuning, not to the test set.
  • Metrics come from the outputs saved in the notebook; the experiment was not rerun for this portfolio.

06Data engineering and analysis

Revenue and delivery times

ELT pipeline to analyze revenue and delivery times, segment by state and category and visualize business trends.

  • Python
  • Pandas
  • SQLite
  • SQL
View repository
Technical approach
  • Extraction, loading and transformation of data for querying in SQLite with SQL.
  • Analysis of revenue and delivery times segmented by state and category, with visualization of the results.

03 Services

How can I help you?

I work remotely with teams and clients in Argentina and abroad, in software development and machine learning.

Software

  • Custom internal tools

    Web systems to manage inventory, stock, loans or other processes that live in spreadsheets today, with a database and access from the browser.

    Example: Inventory System

  • Document automation

    Records, reports or forms generated as DOCX and PDF from data entered on the web, with no copy and paste.

    Example: Records Generator

  • Websites and web applications

    Landing pages and dashboards in React to present a business or organize clients and processes.

    Example: Horizonte Global

ML & Data

  • Data analysis

    Cleaning, exploring and querying your data to understand sales, timelines or business trends.

    Example: Revenue and delivery times

  • Predictive models

    Model comparison and validation metrics to estimate risks or outcomes from historical data.

    Example: Home Credit

  • Integrating models into your system

    Models served through an API to classify images or data inside an application.

    Example: Image Classifier
Tell me about your project

04 Profile

Just as comfortable building an API as training a model.

I’m Franco, a software developer. I work on two fronts with equal dedication: web and backend development, and machine learning.

On the web side, I build APIs, internal tools and interfaces with Go, Python and React, and I take care of how they’re deployed with Docker and Linux.

In machine learning, trained at AnyoneAI, I analyze data, train and compare models with Scikit-learn and LightGBM, and integrate models into services with FastAPI and Redis.

I can join a team from either side, and I make the most of projects that need both.

See my GitHub profile
Role
Web, backend and ML development
Main languages
Go · Python · JavaScript
Location
Chaco, Argentina · remote work
Languages
Native Spanish · advanced English
Available for
Jobs and freelance, Argentina and abroad

Experience and education

  1. Internship

    Software development

    Government House of the Province of Chaco · since April 2026

    Internal tools in Go: the Inventory System with SQL Server and the Records Generator, deployed with Docker on Proxmox.

  2. Freelance

    Web development

    Client projects

    Horizonte Global: multilingual landing page and CRM in React for a travel agency.

  3. Training

    Machine Learning

    AnyoneAI

    Image Classifier, Home Credit and an ELT pipeline for revenue and deliveries.

  4. Studies

    University Programming Technician

    UTN · Facultad Regional Resistencia

    In progress since 2022.

Tools and where they’re used

Backend
Go · Python · FastAPI
Inventory, Records, Classifier
Frontend
React · TypeScript · JavaScript · HTML · CSS
Horizonte Global, Records, Income Manager
Data
SQL Server · PostgreSQL · SQLite · SQL · Redis · Excel / excelize
Revenue, Classifier, Inventory
Infrastructure
Docker · Docker Compose · Linux · Proxmox
Records, Classifier
Data & ML
Pandas · Scikit-learn · TensorFlow / Keras · LightGBM
Home Credit, Classifier
Also worked with
Django / DRF · PHP / Laravel · Java · C · C#

05 Contact

Have a position to fill or a project in mind?

I’m available for job opportunities and freelance projects, in Argentina and abroad. Reach out through whichever channel suits you best.

Project screenshot

Real application interface · Sample data