Julián Osorio Mora

Applied AI Specialist · Early Adopter

I bring agentic AI into an organization's processes, together with its teams: I build the integrations and automations, define what the machine does and what the human does, and leave it running, reliable and well governed.

  • 20 years

    in technology, since 2006

  • 4 banks

    on Grupo Aval's core banking platform, 2012 to 2018

  • Since 2023

    with AI: from my first GPTs to working with agents

Download my CV

Medellín, Colombia · Remote, or hybrid in Medellín

Email me
and let's
talk.

01 / What I've built

Four projects, from core banking to AI agents.

  1. 01 · 2026 – present

    Algo Sentido: a studio run by AI agents

    A web design and development studio. I designed the system AI agents (Claude Code) use to build and run it, under rules, tests and human approval: agentic AI across the whole software lifecycle, with governance.

    • I designed the harness every session runs in: 27 rules loaded by the file being touched, 96 versioned memories, 13 hooks that block what shouldn't happen, and over 400 test cases that run before every push.
    • I built a night agent: every day it cross-checks the repository with five connectors (Gmail, Calendar, Cal.com, Fathom and Meta Ads) and leaves up to seven proposals in an inbox. None runs without my approval, and its guard blocks anything it doesn't recognize.
    • I built the client platform in Next.js, TypeScript, Vercel and Neon: a private portal per client, PayPal payments and automatic receipts through the Gmail API.
    • Releases publish themselves: each branch leaves its note, and a GitHub Action works out the version, writes the changelog and tags it.

    • 1,685

      commits in seven months

    • 317

      self-published releases

    • 400+

      test cases before every push

    Claude Code · MCP · GitHub Actions · Next.js · TypeScript · Vercel · Neon

  2. 02 · 2024 – present

    Interfaz LGC: a subscription app, built end to end with AI

    interfazlgc.com ↗

    A subscription web app for a study community. With AI, I handle everything it takes: design, development, payments and operations.

    • I built an AI router across OpenAI, Anthropic and OpenRouter that switches provider when one fails and tracks tokens and cost per user.
    • I added audio transcription, automatic topics and full-text search over its archive, synced with YouTube and Telegram.
    • I integrated payments with Polar and verified webhooks, and covered it with automated tests and continuous integration on GitHub Actions.

    • 2,483

      commits since December 2024

    • 390

      releases

    • 2,269

      automated tests

    Flask · PostgreSQL · Render · GitHub Actions · OpenAI · Anthropic · Polar

  3. 03 · 2025 – 2026

    AS Tools: AI content tools and their MCP server

    The code, on GitHub ↗

    Tools for working on content with AI: transcription, idea analysis with LLMs, and cutting clips and reels.

    • I built a remote MCP server with OAuth and dynamic client registration, so the tools can be used from claude.ai (May 2026).
    • I encrypted secrets with AES-256-GCM and set up magic-link sign-in.
    • I deployed it on Render, with Docker.

    • 183

      commits in 16 months

    React · Vite · Express · SQLite · Docker · Render · MCP

  4. 04 · 2012 – 2018

    Grupo Aval's core banking: four banks, from QA to production

    The core banking platform used by Grupo Aval's four banks in Colombia: Banco de Bogotá, Banco de Occidente, Banco Popular and Banco AV Villas. Six years, with Excelsys and with Solem.

    • With Excelsys, as head of its department in Colombia, I was in charge of the QA and production infrastructure of all four banks, the ICBS implementation and the migration of their online banking.
    • With Solem, I set up continuous integration and delivery for the development team: Jenkins pipelines with GitLab and GitHub, artifacts in Nexus, scripts in Bash, Python and Groovy, and containers with Docker and Kubernetes on Rancher.
    • I was accountable for every release of the product, from QA to production, as the client's direct technical contact, including the integration between its Oracle and IBM environments and their monitoring.

    • 4

      banks on the same core

    • 6

      years, 2012 to 2018

    Jenkins · GitLab · Nexus · Docker · Kubernetes · Rancher · WebLogic · Ansible · Oracle · IBM · Zabbix · Prometheus · Grafana

Figures as of November 2026.

02 / Where I come from

From tech support to banking, and from banking to AI.

Julián Osorio Mora

I started in tech support and networking. At WOG, a banking software company, I went from support to running its IT platform, with four promotions in four years.

From 2012 to 2018 I worked on Grupo Aval's core banking platform, the one its four banks in Colombia run on: Banco de Bogotá, Banco de Occidente, Banco Popular and Banco AV Villas. With Excelsys, as head of its department in Colombia, I was in charge of the QA and production infrastructure of all four banks. With Solem, I set up continuous integration and delivery for the development team, and I was accountable for every release, from QA to production, as the client's direct technical contact. That's where I learned DevOps end to end: its cycles, its processes, its governance and working with teams.

Then I brought it to other organizations as a consultant: WOG's containerization, BBVA Seguros' deployments with VASS, and Liberty Seguros' pipelines with Amaris.

Since 2022 I've been building on my own, and I adopted AI from its first tools: by 2023 I was making my own GPTs. Since 2024, with AI, I've handled everything Interfaz LGC takes, a subscription web app for a study community: design, development, payments and operations. So far, 2,483 commits, 2,269 automated tests and a router that switches AI provider when one fails. Today I run my projects with AI agents, under rules, tests and human approval: what I learned in banking, applied to AI.

That's why I know how an organization that delivers technology works, and how it changes when AI comes in.

My job is to implement it so it truly makes sense: so it improves each area and revenue per employee, and keeps running, with its teams and with governance.

Career

  1. 2022 – present

    Independent

    Technology and AI consultant

    Interfaz LGC, AS Tools and Algo Sentido; the technology behind brands like Reset & Rise and Atelier Coworking, and AI proofs of concept.

  2. 2021 – 2022

    Amaris Consulting

    Consultant

    For Liberty Seguros in Colombia, and for SMU (Unimarc) and Cencosud in Chile: DevOps, pipelines and Kubernetes.

  3. 2020 – 2021

    VASS

    DevOps Consultant

    For BBVA Seguros: moved its web projects to Maven, with their packaging, deployments and logs.

  4. 2019 – 2021

    WOG

    Infrastructure and DevOps consultant, under contract

    Virtualized its environments, containerized its applications with Rancher and recovered its Rancher on AWS.

  5. 2014 – 2018

    Solem

    Infrastructure Engineer

    Grupo Aval's core banking: continuous integration and delivery, and every release from QA to production.

  6. 2014

    Tata Consultancy Services

    Architect

  7. 2012 – 2014

    Excelsys

    Infrastructure Implementation Engineer

    Grupo Aval's core banking: head of the department in Colombia, in charge of QA and production infrastructure.

  8. 2008 – 2012

    WOG

    Platform Administrator

    From level 1 tech support to running its IT platform, with four promotions.

  9. 2007 – 2008

    Internacional de Negocios

    Systems Assistant

  10. 2006 – 2007

    Goldentech

    Network Technical Support

Education
  • Systems Technologist · Universidad ECCI
  • Nine semesters of Systems Engineering · Universidad San Martín and Universidad ECCI
  • Systems Technician · Instituto INSCAP

03 / How I work

What belongs to the machine, and what to the human?

When I bring AI into a process, I start by understanding it end to end. Then I separate the tasks that belong to the machine from the ones that belong to people, and I look for how to make them converge.

The technical side has gotten cheaper. Where the cost has really gone up is in judgment.

So it's not enough for it to work: it has to keep working, and it has to be governed. I look for a harness where we keep learning and stabilizing all of it.

My career has moved through the technical side, architecture, execution, creation, strategy and integration, and through connecting the dots, systems thinking and process. One example of each:

  • 01

    The technical side

    I started in tech support and networking, and at WOG I ended up running its IT platform. Today, the code of my apps, written with AI.

  • 02

    Architecture

    The QA and production infrastructure of Grupo Aval's four banks. Today, my apps': Flask and PostgreSQL with an AI router; Next.js, Vercel and Neon.

  • 03

    Execution

    Every core banking release, from QA to production, as the client's technical contact. Recovering WOG's Rancher on AWS. Today, 390 releases of Interfaz LGC.

  • 04

    Creation

    My GPTs and an AI course for entrepreneurs, in 2023. Interfaz LGC, end to end with AI. AS Tools and its MCP server for claude.ai.

  • 05

    Strategy

    The containerization I proposed to WOG. The technology and strategy behind brands like Reset & Rise. AI proofs of concept for a Colombian family compensation fund.

  • 06

    Integration

    Continuous integration and delivery for core banking; Liberty Seguros' pipelines. Today, PayPal, Polar, Cal.com, the Gmail API and MCP servers.

  • 07

    Connecting the dots

    What I learned in banking, applied to AI: my agents work under rules, tests and human approval.

  • 08

    Systems thinking

    A harness of rules, memories, hooks and tests that work together, and an agent that learns from my answers: three yeses in a row propose automating; three noes, adjusting.

  • 09

    Process

    DevOps end to end: governance, cycles and processes. WOG's containerization, with its method: an inventory, interviews with each team and a board.

04 / What I'm looking for

I want to bring agentic AI into a company.

It excites me, and it's a good time to do it: to learn, and to see its challenges.

I've worked as an employee, as a consultant and on my own, always with the same mechanics: I understand the process end to end, build what makes it work, and leave it reliable and governed. It works whatever the subject, the department, the process or the flow, and it's one of my strengths. What I did on my own are projects I built or took part in, and they shaped this path: that's the experience that makes sense to bring into a company today.

Build or lead? Both, and knowing when each one makes more sense. I think through the architecture and ship an MVP with AI as an ally, to make things real and test them fast, and I know how to delegate and orchestrate the implementation. I can be part of a well-organized team and, from my role, do things with the highest possible quality. I see the process, the flow and the system, and I keep an eye on everything lining up.

Roles
Applied AI, and AI adoption across an organization (AI Enablement, AI Transformation).
Where
Remote from Medellín, Colombia, or hybrid or on-site in Medellín.
Engagement
Employment or a services contract.
Start
Right away.
Languages
Spanish, native. English: technical and reading; basic conversation.

05 / Contact

Email me and let's talk.

info@julianosoriom.com