Skip to content
Jorge Sierra

Open to new challenges

From chaos
to architecture.

I'm Jorge Sierra, AI Engineer & Senior Full-Stack. I turn manual processes and systems that don't talk to each other into agents, automations and APIs that scale.

Five-layer architecture: INPUTS → ORCHESTRATION → AI AGENTS → APIs → DATA

See case studiesBook a call

n8n · LangChain · RAG · NestJS · Next.js · Supabase

02 · Case studies

Real problems, visible architecture.

Five systems, from the most finished to the one on its way. Swipe to pick one.

Case 01Finished · live demo

From notebooks and Excel to a real-time collections platform

Paga Diario

Demo (opens in a new tab)Code (opens in a new tab)

Data flow

  1. FIELDCollectorsGPS collection routes
  2. APINode.js + ExpressJWT · sessions
  3. QUEUESBullMQ + Redisbackground jobs
  4. DATAPostgreSQL 16Prisma ORM
  5. OUTPUTREAL TIMELive dashboardSocket.io · Telegram alerts

The problem

Daily-collection microloan portfolios are usually kept in notebooks and Excel sheets. Paga Diario automates payment recording, late-fee calculation and reports, and organizes the collectors' routes.

Key decisions

  1. Late fees under Colombian regulationThe calculation is automatic and follows the regulation instead of relying on manual math.
  2. Real time without blocking the APIBullMQ on Redis and Socket.io for the live dashboard. Redis is optional: basic operation works without it.
  3. Tested and deployedTests with Jest and Supertest, Docker containers and deployment on Fly.io.

Outcome

  • Live demo with public credentials.
  • Complete modules: loans, payments, collectors, expenses, bulk import, reports and Telegram alerts.
  • Dato pendiente: [Next step: an AI assistant that prioritizes the day's route and flags clients at risk of late payment]
  • Node.js
  • Express
  • PostgreSQL
  • Prisma
  • BullMQ
  • Socket.io
  • Fly.io

Do you have a process that still runs on spreadsheets and manual work?

Ask my agentSee GitHub (opens in a new tab)

03 · Experience

Over five years turning processes into systems.

Backend, frontend, cloud and now AI agents. Filter by area to see where I used each one.

  1. 2025 — PresentCurrent

    AI & Automation Engineer

    SOAINT

    • I lead a team of 2 to 3 developers.
    • I design and build AI automation flows in n8n, with language-model agents that replace repetitive manual processes.
    • I build and maintain AI pipelines with LangChain and LangGraph that connect language models to internal tools, APIs and data sources.
    • n8n
    • LangChain
    • LangGraph
    • LangSmith
    • RAG
    • Pinecone
    • Python
    • FastAPI

  2. 2023 — 2025

    Full Stack Developer

    IX Colombia

    • I built administrative modules for patient management on a healthcare platform.
    • I achieved 100% traceability in patient flows and cut operational errors by 20% with validations and improvements to the interface logic.
    • I started on the frontend with Vue and Quasar and then moved to the backend.
    • Product used by clients in Colombia, Peru and Argentina.
    • React
    • Vue
    • Quasar
    • TypeScript
    • FastAPI
    • GraphQL
    • Redis
    • GitLab CI
    • Terraform

  3. 2019 — 2023

    Full Stack Developer

    SENA

    • Backend with Python, Django and Django REST Framework, applying patterns such as Singleton and Factory.
    • GCP services: Compute Engine, Cloud Functions, Cloud Run, GKE, Pub/Sub, BigQuery and App Engine.
    • Python
    • Django
    • DRF
    • NestJS
    • PostgreSQL
    • GCP
    • BigQuery

04 · Contact

Tell me about your business problem.

Write to me and watch what happens to your message: it is the same kind of automation I build for others.

What brings you here?

What happens when you send

DEMO · IN PRODUCTION IT RUNS ON n8n

  1. Webhook receivedWaiting
  2. AI classificationWaiting
  3. Saved to the CRMWaiting
  4. Jorge notifiedWaiting
  5. Automatic replyWaiting
$ waiting for a message…

Medellín, Colombia · Remote or hybrid · I reply within 24 h