● Open to mid-senior backend / full-stack rolesBusco roles mid-senior de backend o full-stack

Jordano
Pernia

Full-stack engineer building production LLM systems. I take products from empty repo to daily use: architecture, backend, frontend, deploy, then iterate with real users.Ingeniero full-stack. Construyo sistemas con LLM que ya están en producción. Llevo cada producto desde el repositorio vacío hasta que la gente lo usa a diario: arquitectura, backend, frontend, despliegue y mejoras con usuarios reales.

$ whoami
Software EngineerSoftware Engineer
$ location
Táchira, Venezuela · Remote · GMT-4Táchira, Venezuela · Remoto · GMT-4
$ stack
Python · TypeScript · Go · LLM systems
$ status
Available from Oct 2026Disponible desde octubre de 2026
$ contact
jordano470@gmail.com

AboutSobre mí

Full-stack software engineer with 4 years building web products and 1+ year shipping production LLM systems. I architected and built the KonEmpleo work environment, four portals and Arka, an AI recruiting agent, from an empty repo to daily use by recruiters and their clients.Ingeniero de software full-stack con cuatro años construyendo productos web y más de un año llevando sistemas con LLM a producción. Diseñé y construí el entorno de trabajo de KonEmpleo, cuatro portales y Arka, un agente de IA para reclutamiento, desde un repositorio vacío hasta convertirse en la herramienta diaria de reclutadores y sus clientes.

Comfortable across Python, TypeScript and Go; strongest at turning a vague business need into a working system, then iterating on it with real users.Me muevo con soltura entre Python, TypeScript y Go. Donde más aporto es en convertir una necesidad de negocio poco definida en un sistema que funciona, y luego mejorarlo con usuarios reales.

B.Eng. Computer Engineering, UNET · index 7.78 / 9Ingeniería en Informática, UNET · índice 7,78 / 9

ExperienceExperiencia

From agencies and freelance to building AI products from scratch.De agencias y trabajo freelance a crear productos con IA desde cero.

Seika AISeika AI

Jan 2026 – Oct 2026Ene 2026 – Oct 2026
Software EngineerSoftware EngineerBogotá, Colombia (remote)Bogotá, Colombia (remoto)
  • Architected and built the KonEmpleo work environment from empty repo to daily use on a 2-person team: four Next.js portals and Arka, an AI recruiting agent, on a Django REST + Celery backend with 113 endpoints, 57 models and ~1,900 automated tests.
  • Built semantic candidate search on pgvector (HyDE query expansion, keyword and role boosts) and Arka, an AI recruiting agent built on Pydantic AI with tool-calling, prompt caching and cost telemetry.
  • Improved Equifinanzas' Claude-vision document-processing pipeline, which files accounting records into Google Drive and Sheets, cutting per-document handling from minutes to seconds.
  • Diseñé y construí el entorno de trabajo de KonEmpleo, desde el repositorio vacío hasta su uso diario, en un equipo de dos personas: cuatro portales en Next.js y Arka, un agente de IA para reclutamiento, sobre un backend en Django REST y Celery con 113 endpoints, 57 modelos y cerca de 1.900 pruebas automatizadas.
  • Desarrollé la búsqueda semántica de candidatos sobre pgvector (expansión de consultas con HyDE y ponderación por palabras clave y cargo) y Arka, un agente de IA para reclutamiento construido con Pydantic AI, con tool-calling, prompt caching y telemetría de costos.
  • Mejoré el pipeline de procesamiento de documentos de Equifinanzas, basado en Claude vision, que archiva registros contables en Google Drive y Sheets: el manejo de cada documento pasó de minutos a segundos.

UNET – Office of ResearchUNET – Decanato de Investigación

Nov 2024 – Dec 2025Nov 2024 – Dic 2025
Software EngineerSoftware EngineerTáchira, VenezuelaTáchira, Venezuela
  • Built RutUNET, a real-time fleet tracking system: Go backend with live location feed, React dashboard and React Native app.
  • Added usage statistics and analytics on rider behavior for university transport planning.
  • Delivered internal tools for the campus digitalization program.
  • Construí RutUNET, un sistema de seguimiento de flota en tiempo real: backend en Go con ubicación en vivo, panel web en React y app móvil en React Native.
  • Incorporé estadísticas de uso y analítica del comportamiento de los pasajeros para apoyar la planificación del transporte universitario.
  • Desarrollé herramientas internas para el programa de digitalización del campus.

FreelanceFreelance

Jun 2024 – Dec 2025Jun 2024 – Dic 2025
Software EngineerSoftware EngineerRemoteRemoto
  • 6+ websites and internal management systems for small businesses, with payments and transactional email integrated.
  • Más de seis sitios web y sistemas de gestión interna para pequeños negocios, con pasarela de pagos y correo transaccional integrados.

Sistemas del CaribeSistemas del Caribe

Nov 2023 – May 2024Nov 2023 – May 2024
Software EngineerSoftware EngineerQuintana Roo, MexicoQuintana Roo, México
  • Web applications from scratch for local businesses, scoping requirements directly with clients.
  • Aplicaciones web desde cero para negocios locales, levantando los requisitos directamente con cada cliente.

Bidaia TravelBidaia Travel

Aug 2022 – May 2024Ago 2022 – May 2024
Software EngineerSoftware EngineerCalifornia, US (remote)California, EE. UU. (remoto)
  • Booking website with Google Maps API, Stripe payments and a custom REST API, built from Figma designs.
  • Sitio de reservas con la API de Google Maps, pagos con Stripe y una API REST propia, desarrollado a partir de diseños en Figma.

Selected ProjectsProyectos destacados

Production systems I architected and shipped, with real users. Every one links to the live site.Sistemas en producción que diseñé y entregué, con usuarios reales. Cada uno enlaza al sitio en vivo.

Screenshot of KonEmpleo
01AI · RecruitingIA · Reclutamiento

KonEmpleoKonEmpleo

The KonEmpleo work environment, built from scratch for a recruiting firm: four Next.js portals for AI talent search and outsourcing, and Arka, an AI recruiting agent with tool-calling. Django REST + Celery backend with semantic candidate search on pgvector. Developed through constant meetings with the KonEmpleo team; in daily use by recruiters and their clients.El entorno de trabajo de KonEmpleo, construido desde cero para una firma de reclutamiento: cuatro portales en Next.js para búsqueda de talento y outsourcing con IA, y Arka, un agente de IA para reclutamiento con tool-calling. Backend en Django REST y Celery, con búsqueda semántica de candidatos sobre pgvector. Se desarrolló en reuniones constantes con el equipo de KonEmpleo y hoy lo usan a diario reclutadores y sus clientes.

113 endpoints~1,900 tests2-person team
113 endpoints~1.900 pruebasEquipo de dos
Visit siteVer sitio ↗
Screenshot of Equifinanzas
02AccountingContabilidad

EquifinanzasEquifinanzas

Accounting-firm platform where my main contribution was improving the document-processing pipeline. Claude vision classifies each record and extracts structured fields from PDFs, scans and phone photos, then files them into Google Drive and Sheets automatically.Plataforma de una firma contable en la que mi aporte principal fue mejorar el pipeline de procesamiento de documentos. Claude vision clasifica cada registro, extrae los campos estructurados de PDFs, escaneos y fotos tomadas con el celular, y los archiva automáticamente en Google Drive y Sheets.

Minutes → seconds per docRuns daily
De minutos a segundos por documentoEn uso diario
Visit siteVer sitio ↗
Screenshot of Ophthalmology Service
03HealthcareSalud

Ophthalmology ServiceServicio de Oftalmología

Surgical notes system for the Servicio de Oftalmología, Hospital Central de San Cristóbal, designed in person with the service's physicians. Cut the time to write each operative note from 20 minutes to about 3.Sistema de notas quirúrgicas para el Servicio de Oftalmología del Hospital Central de San Cristóbal, diseñado en reuniones presenciales con los médicos del servicio. Redactar cada nota operatoria pasó de tomar 20 minutos a unos 3.

20 min → ~3 min per note
De 20 min a ~3 min por nota
Visit siteVer sitio ↗
Screenshot of RutUNET
04TransportationTransporte

RutUNETRutUNET

Real-time transportation management and tracking system for the UNET university fleet. Go backend with a live location feed, React web dashboard and React Native app for riders and drivers, with usage statistics and analytics on rider behavior. Currently being adopted by UNET's administration.Sistema de gestión y seguimiento en tiempo real de la flota de transporte de la UNET. Backend en Go con transmisión de ubicación en vivo, panel web en React y app en React Native para pasajeros y conductores, con estadísticas de uso y analítica del comportamiento de los pasajeros. La administración de la UNET está en proceso de adoptarlo.

Go · React · React NativeBeing adopted by UNET
Go · React · React NativeEn adopción por la UNET
Visit siteVer sitio ↗
Screenshot of Pensum UNET
05EducationEducación

Pensum UNETPensum UNET

Maps every course in a UNET degree and its prerequisites as a navigable graph, lets students calculate their grades and academic index, and recommends the optimal courses for next semester with a DFS-based planner.Convierte cada carrera de la UNET y sus prelaciones en un grafo navegable. Permite a los estudiantes calcular sus notas y su índice académico, y recomienda las materias óptimas para el próximo semestre con un planificador basado en DFS.

Used by every student on campusDFS-based planner
Lo usa todo el campus de la UNETPlanificador basado en DFS
Visit siteVer sitio ↗

How I workCómo trabajo

01

Requirements firstPrimero los requisitos

Every project starts with the client's requirement list. I turn it into a PRD with use cases and test scenarios before writing code, so scope, edge cases and acceptance criteria are agreed up front.Cada proyecto arranca con la lista de requisitos del cliente. Antes de escribir código la convierto en un PRD con casos de uso y escenarios de prueba, para que el alcance, los casos límite y los criterios de aceptación queden acordados desde el principio.

02

Architecture that growsArquitectura que crece

KonEmpleo began as a monorepo monolith because it was a simple single-purpose site. As it grew I evolved the backend into a multitenant system serving four portals and external clients, without a rewrite.KonEmpleo nació como un monolito en un monorepo, porque al inicio era un sitio sencillo con un solo propósito. A medida que creció, convertí el backend en un sistema multitenant que hoy atiende cuatro portales y clientes externos, sin reescribirlo.

03

AI-assisted, human-reviewedCon IA, revisado por personas

I use Claude Code daily to move faster, and design codebases so that agents and humans can work in them safely: small modules, clear conventions, every diff reviewed.Uso Claude Code a diario para avanzar más rápido, y diseño las bases de código para que agentes y personas puedan trabajar en ellas con seguridad: módulos pequeños, convenciones claras y cada cambio revisado.

SkillsHabilidades

LanguagesLenguajes
  • Python
  • TypeScript / JavaScript
  • Go
  • SQL
  • Python
  • TypeScript / JavaScript
  • Go
  • SQL
BackendBackend
  • Django REST
  • Celery
  • PostgreSQL
  • pgvector
  • REST API design
  • JWT / OTP auth
  • Django REST
  • Celery
  • PostgreSQL
  • pgvector
  • Diseño de APIs REST
  • Autenticación JWT / OTP
FrontendFrontend
  • React
  • Next.js
  • Astro
  • Tailwind CSS
  • Figma
  • React
  • Next.js
  • Astro
  • Tailwind CSS
  • Figma
AI / LLMIA / LLM
  • Anthropic & OpenAI APIs
  • Pydantic AI
  • Tool-calling agents
  • Embeddings & semantic search
  • Document extraction / OCR
  • APIs Anthropic y OpenAI
  • Pydantic AI
  • Agentes con tool-calling
  • Embeddings y búsqueda semántica
  • Extracción de documentos / OCR
InfraInfra
  • Docker
  • Railway
  • Git / GitHub
  • Claude Code
  • Docker
  • Railway
  • Git / GitHub
  • Claude Code
SpokenIdiomas
  • Spanish (native)
  • English (C1)
  • Español (nativo)
  • Inglés (C1)

Let's talkHablemos

jordano470@gmail.com