AI Developer

Location

Portland, OR (in-office part time)

$90-120K

1-3 Years or equivalent 'proof of work' and side projects

CTO

Most junior roles involve fixing bugs in legacy systems. This isn't one of them.

At Zelody we are rebuilding financial infrastructure from the ground up using an AI-first approach. We are looking for a hungry, "AI-native" Software Engineer who is obsessed with the potential of LLMs and autonomous agents.

You will work directly with our CTO and Principal Architect to turn experimental AI concepts into production-grade financial tools. You'll build and ship AI-driven workflows that connect APIs, transform data, and automate real financial operations.

This is a learn as you go position where your curiosity, work ethic, and drive is as important as your experience and current skillset. You will:

• Prototype:
  Rapidly build "proof of concept" features using AI frameworks
• Test and improve the platform:
  Work alongside senior mentors to learn how to make AI reliable, secure, and compliant in a regulated banking environment.
• Ship Production Code:
  Contribute to our core Type Script/Node.js stack, ensuring our AI agents play nice with our broader infrastructure.

• 1--3 years of professional software engineering experience (Fintech experience is a "nice to have," but your ability to build is what matters).
• A portfolio of "AI Experiments":
  Whether it's a side project, a Git Hub repo, or a complex automation you built for fun---we want to see what you've tried.
• Build & Iterate:
  Develop and refine agentic workflows that handle real-world financial scenarios.
• Understanding of LLMs:
  You understand the basics of how LLMs work (tokens, context windows, temperature, and RAG).
• The "Zelody" Pace:
  You thrive in a startup environment where things move fast and your work has a direct impact on the company's success.

• Languages:
  
  Experience with
  Type Script/Node.js
• Infrastructure:
  Some experience with cloud systems like AWS, event-driven architectures.
• Early Experience with vector databases (Pinecone, Weaviate, etc.).
• Examples:
  You've built something that uses an LLM to "do" something (execute a tool or call an API), not just "say" something.