Product Engineer Intern, Agent Systems

About Mactores Mactores is the agent-native AWS modernization firm. Most enterprise modernization stalls at pilots and handoffs; ours ships, production systems running, legacy retired, outcomes measured. Our delivery is built on Aedeon, the agent platform built by Mactores' founders' sister company, Aedeon Inc. Aedeon absorbs the repetitive 60–70% of engagement work, discovery, mapping, validation, test generation, while forward-deployed engineers own architecture, judgment, and cutover. Mactores is Aedeon's flagship deployment partner: field delivery informs the platform roadmap, and the platform shapes how we deliver. About Aedeon Aedeon is the agent platform for enterprise modernization, built by Aedeon Inc., a separate product company founded by the same team that founded Mactores. Aedeon turns the systems already running a business applications, databases, data platforms, business rules, workflows into governed AI agents, grounded in a persistent Code Intelligence Graph of the customer's own code and verified through behavior-equivalence proof. Aedeon is delivered as a product, not a services engagement, and runs without mandatory forward-deployed engineers. As an intern on Aedeon's Agent Systems team, you'll work alongside the engineers who build the agents that do the actual modernization work: parsers that read legacy estates, extractors that lift business rules out of code, verifiers that produce behavior-equivalence proof. You won't own an agent end to end on day one, but you'll write real Python against the Code Intelligence Graph, under close mentorship, and see how governed autonomy (shadow, supervised, autonomous stages) actually works in production. You don't need prior experience with Bedrock or LangChain. You need to be a curious, careful Python learner who wants to understand agent frameworks deeply rather than treat them as a buzzword.

What you'll do?

• Agent systems development

• Assist in building components of the Aedeon agent fleet parsers, extractors, dependency mappers, test synthesizers under senior engineer guidance.
• Learn how agents operate against the Code Intelligence Graph so every action traces back to a line of code.
• Get exposure to governed autonomy stages (shadow, supervised, autonomous) and the human-in-the-loop controls at each stage.
• Shadow how the team orchestrates foundation models (Anthropic, OpenAI, Google) through the Aedeon Decision Model.

Verification and quality

• Practice writing tests before code, for the small pieces of functionality you own.
• Help build behavior-equivalence checks comparing agent output against expected results.
• Contribute to test cases covering functional flows and edge cases, with mentor review before anything ships.
• Get hands-on with pytest and the team's CI/CD-integrated test suites.

Collaboration and documentation

• Write clear, readable Python and submit pull requests for review.
• Document what you build in Confluence or equivalent clearly enough that a teammate can pick it up without asking you first.

What we're looking for?

• Core Python and fundamentals

• Currently pursuing, or recently completed, a degree in Computer Science, Engineering, or a related field.
• Solid foundation in Python personal projects, internships, or coursework count; no minimum years of professional experience required.
• Basic understanding of APIs and web frameworks (FastAPI, Flask, or similar) is a plus.
• Genuine curiosity about LLMs, agent frameworks, and how software gets verified before it ships.
Engineering mindset
• Willingness to learn testing discipline as a habit, not an afterthought.
• Comfortable asking questions and taking feedback on code quality.
• Available to work with reasonable overlap with US business hours.

You'll be preferred if you've

• Coursework or personal projects using LangChain, AWS Bedrock, or similar agentic AI frameworks.
• Any exposure to Docker, basic Kubernetes concepts, or cloud platforms (AWS preferred).
• Familiarity with graph databases (Neo4j) or Elasticsearch/OpenSearch, even at a conceptual level.