Automation & AI
Workflows, agents and document systems
PythonAI AgentsReport AutomationDocument GenerationValidation Workflows
Alejandro Padilla
Full-Stack Developer
I build and run full-stack systems: FastAPI and Next.js services on PostgreSQL, shipped with Docker, and LLM automation where it cuts real manual work.
Alejandro Padilla
open · part-time / contract
$ about-me
Systems that ship, built for a business reason.
I'm Alejandro Padilla, a full-stack developer and Computer Science student at Universidad Nacional de Loja in Ecuador.
I build and operate systems that real clients depend on: domain modeling, backend services, asynchronous job processing, authentication, automated tests, and deployment. I also apply AI as an engineering tool — LLM-backed automation and agent-orchestrated development workflows — when it removes measurable manual work.
EN B1 · ES native1
How I work
Clear scope before code, short deliveries, documented tests and deployment. AI as a working tool, not a headline.
$ tech-stack
Tools for building and operating reliable software.
Backend & Data
APIs, persistence and data handling
JavaPostgreSQLSQLPrismaMavenCSV Data Processing
Frontend
Interfaces and full-stack apps
HTMLCSSReactNext.jsTailwind CSS
Tools
Daily development workflow
GitGitHubDockerLinux
Delivery
Documented, reproducible software
GitHub ActionsGitHub PagesTestingDocumentationResponsive UI
$ featured-projects
Selected work, from production systems to applied AI.
In Production
Live demo · free tier, first load may take a few seconds01 / projectPublic source
YOLO Complexity Lab
Computer-vision lab that benchmarks YOLO11n against Faster R-CNN and SSDLite on the same input — latency, FPS, GFLOPs, parameters — with a live webcam mode and custom weights.
- ▸ Live webcam mode: per-frame latency, FPS and detections through YOLO11n.
- ▸ Two-stage vs one-stage vs YOLO benchmark with mean/p95 latency, a winners chart and CSV export.
- ▸ Deployed on Streamlit Cloud with GitHub Actions CI.
PythonPyTorchStreamlitYOLO11Benchmarking
Live client site · custom domain02 / projectClient project
El Horno del Pingüino
Artisanal bakery in Loja: a storefront with catalog, business orders, FAQ and a single WhatsApp ordering flow, plus an admin panel where the owners keep products, banners and FAQ up to date themselves.
- ▸ Built the full responsive storefront: catalog, flavors, bulk orders, FAQ and a three-step ordering guide.
- ▸ The catalog comes from a Next.js + Prisma admin panel; a nightly rebuild publishes the owners' changes with no code edits.
- ▸ Every section funnels into one WhatsApp CTA, matching how the business closes sales.
AstroNext.jsPrismaPostgreSQLWhatsApp CTA
Academic & Open Source
Source and setup documented01 / projectPublic source
academic-report-flow
Python toolkit that turns Markdown and YAML sources into university-grade PDF reports: a LaTeX route with bibliography and validation gates, a WeasyPrint route for fast previews, IEEE reference checks and reproducible figures.
- ▸ Two build routes: Markdown → LaTeX → PDF with bibliography and validation gates, and Markdown → HTML → PDF for previews.
- ▸ Validates report structure and IEEE-style references before anything ships.
- ▸ Generates Mermaid, Vega-Lite and ECharts figures as reproducible assets.
PythonLaTeXWeasyPrintPlaywrightValidation
Source and deployment documented02 / projectPublic full-stack project
AlejandroTatum.github.io
Bilingual portfolio with a terminal-style dev-mode UI: one typed content model for EN/ES, GSAP ScrollTrigger and Lenis motion with reduced-motion support, static export to GitHub Pages.
- ▸ Supports English and Spanish content from one typed content model.
- ▸ Uses a responsive interface with accessible semantic sections.
- ▸ Builds and deploys as a static site through GitHub Pages.
Next.jsTypeScriptReactTailwind CSSGitHub Pages
$ run hospital-appointment-inventory
Source and CLI demo documented
03 / projectPublic Java project
hospital-appointment-inventory
Documented Java CLI that models appointment, patient and inventory workflows over hospital-style CSV datasets.
- ▸ Loads and processes hospital-style CSV datasets.
- ▸ Compares sorting and search strategies over practical scenarios.
- ▸ Includes a documented CLI demo and Maven Wrapper setup.
Java 21MavenApache Commons CSVJUnit 5Data StructuresSearch AlgorithmsSorting Algorithms
$ run siged
Source and setup documented
04 / projectPublic academic project
siged
Coursework implementation of a SIGED-style school management system: Django backend and React client, built against a real reference system.
- ▸ Implements an education management domain following a SIGED-style reference system.
- ▸ Separates a Django backend from a React client interface.
- ▸ Built as graded coursework against a real reference system.
DjangoReactPythonEducation Management
$ run hmm-speech-recognition-research
Research document and references documented
05 / projectPublic academic research
hmm-speech-recognition-research
Literature review of HMM-based speech recognition and its evolution toward hybrid and deep models, typeset as a reproducible LaTeX document.
- ▸ Reviews the foundations of Hidden Markov Models for speech recognition.
- ▸ Traces the evolution from classic HMM pipelines to hybrid and deep models.
- ▸ Typeset as a reproducible LaTeX research document.
ResearchHMMSpeech RecognitionLaTeX
$ run search-algorithms-java
Source and setup documented
06 / projectPublic Java academic project
search-algorithms-java
Java academic project implementing sequential and binary search over arrays and linked lists using CSV datasets.
- ▸ Implements sequential and binary search over arrays and linked lists.
- ▸ Loads CSV datasets to compare search behavior on real data.
- ▸ Structured as a documented academic exercise.
JavaData StructuresSearch AlgorithmsCSV
$ contact
Let's build something that ships.
A system to build, or a team that needs another engineer? Tell me the problem. You'll get a read on scope and the first steps.