About

Hi there! I'm Elisban, and I love building robust systems. I'm a software engineer specialized in backend architecture, data pipelines, and applied machine learning. I take pride in turning messy, large-scale data into clean, deterministic software that people can rely on. I enjoy working most at the intersection of algorithmic depth and practical systems engineering where solid mathematical foundations meet high-performance code.

Currently, I architect high-throughput backend services and serverless MLOps workflows. I'm also the creator of production platforms like brevio.link (a high-speed URL platform built with island architecture) and platanito.dev (an edge-distributed editorial platform). I care deeply about latency budgets, reproducible pipelines, and clean relational database modeling.

Over the past few years, I've built across a wide spectrum of environments from engineering citizen-facing mobile applications for the Municipality of Coasa and core institutional APIs at UNAP, to architecting causal Bayesian networks for Cybersecurity AI and building high-performance systems. These experiences taught me that the best systems are not the most complex ones, but the simplest abstractions that solve the problem reliably.

When I'm not profiling database queries or training models, you can usually find me tackling algorithmic puzzles in competitive programming contests, exploring the high-altitude landscapes of Puno, or diving into computer science papers on information theory.

Experience

  1. Apr. 2026 — Jul. 2026

    Oficina de Tecnologías de Información (OTI) · UNAP

    Engineered and optimized core institutional backend services, high-concurrency RESTful APIs, and database migrations in PostgreSQL for campus-wide administrative platforms and digital student services.

    • FastAPI
    • PostgreSQL
    • TypeScript
    • RESTful APIs
    • Docker
    • Git
  2. Feb. 2025 — Mar. 2025

    Mobile & Systems Developer

    Municipalidad Distrital de Coasa

    Architected and deployed a native Android citizen-facing mobile application using Kotlin and Jetpack Compose, modernizing municipal workflows, citizen requests, and REST backend integrations.

    • Kotlin
    • Jetpack Compose
    • Android SDK
    • REST APIs
    • Mobile Architecture
View Full Résumé

Projects

  • Adaptive Learning Recommendation Engine & MLOps Pipeline

    A smart system that learns user preferences over time to recommend movies they will actually enjoy.

    An end-to-end adaptive movie recommendation engine combining SVD matrix factorization, K-Means clustering, and a Q-Learning reinforcement agent to dynamically balance exploitation vs. exploration. Includes automated serverless MLOps CI/CD/CT with Databricks MLflow and GitHub Actions.

    RMSE Reduction: 0.842
    • Python
    • FastAPI
    • SVD
    • Q-Learning
    • Databricks MLflow
    • GitHub Actions
    • Docker
    • Render
    • uv
    Adaptive Learning Recommendation Engine & MLOps Pipeline Preview

  • Multivariate Vocational Recommender System

    An AI chatbot and psychological test (validated by the Universidad Nacional del Altiplano) that uses natural language and the Mahalanobis algorithm to recommend university careers.

    A vocational recommendation system featuring a situational psychological test validated by professionals from the Universidad Nacional del Altiplano. It includes an AI chatbot that interacts with students in natural language (via API or local AI) and uses the Mahalanobis algorithm to perfectly match them with their ideal university degree.

    Profile Dimensions: 16D
    • FastAPI
    • Next.js 16
    • SciPy
    • Mahalanobis Distance
    • Mem0 Cloud
    • Python
    • pnpm
    Multivariate Vocational Recommender System Preview

  • High-Performance NCD & MST Data Analysis Pipeline

    A fast program that automatically groups massive amounts of files by calculating how similar they are to each other.

    A high-performance C++17 and Python pipeline for analyzing educational data using Normalized Compression Distance (NCD) and Minimum Spanning Trees (MST). Implemented Kruskal's and Prim's graph algorithms in C++ for optimal tree extraction, achieving millisecond processing times on 18,000+ records.

    Dataset Scale: 18,000+ records
    • C++17
    • Python 3
    • CMake
    • Graph Algorithms
    • Data Analysis
    • Information Theory
    • Kruskal MST
    High-Performance NCD & MST Data Analysis Pipeline Preview

  • Brevio.link - Enterprise-Grade URL Shortening Platform

    A fast and secure web platform for shortening long web addresses and tracking how many people click them.

    Architected a high-speed enterprise URL shortening platform (brevio.link) leveraging Astro SSR multi-framework island architecture with React 19 and Svelte 5. Integrated PostgreSQL with Prisma 7 Driver Adapters, Zero-Trust Astro Actions, and Cloudflare Turnstile antibot protection.

    Framework Architecture: Astro SSR Islands
    • Astro
    • React 19
    • Svelte 5
    • TypeScript
    • PostgreSQL
    • Prisma 7
    • Lucia Auth
    • Cloudflare Turnstile
    • Tailwind CSS v4
    Brevio.link - Enterprise-Grade URL Shortening Platform Preview

  • toBlog -High-Performance Editorial Publishing Platform

    A high-speed publishing platform that allows writers to create, manage, and share editorial articles globally.

    Architected a modern editorial publishing platform (platanito.dev) using Astro SSR, React 19, and an edge distributed database with Turso (LibSQL) and Drizzle ORM. Engineered an anti-abuse metric system with 24h sliding-window unique view tracking, DOM reading engagement observers, and Cloudflare R2 storage.

    Edge Database: Turso (LibSQL)
    • Astro
    • React 19
    • TypeScript
    • Turso LibSQL
    • Drizzle ORM
    • Lucia Auth
    • Cloudflare R2
    • Tailwind CSS v4
    toBlog -High-Performance Editorial Publishing Platform Preview
View Full Project Archive