Full stack developer for AI driven software development project (33589-1)

SourcePower; Rotterdam; On-site; EUR 70,000 - 120,000; Full time

Job summary

SourcePower seeks an experienced full-stack developer to lead an AI-driven debt management platform rebuild. You will guide AI agents for enterprise-grade delivery across backend, frontend and cloud infra, ensuring guardrails and standards are met.

AWS and Kubernetes expertise are essential in a fast-paced project led by Manoj and Edwin, with tight timelines. Office days are Tuesday and Thursday in Rotterdam, with fully English-speaking teams.

Qualifications

• Experience with AI-driven software development and AI agents.
• Enterprise-grade full-stack development experience.
• Ability to direct AI agents and set guardrails/standards.

Responsibilities

• Lead and instruct AI agents to design and implement enterprise software features.
• Ensure architectural integrity across backend, frontend, and infra.
• Collaborate with stakeholders to align AI-driven approaches with business goals.
• Maintain security and compliance during accelerated delivery timelines.

Skills

Full-stack

AI-guided dev

AWS Kubernetes

System architecture

Security awareness

English communication

Tools

AWS

Kubernetes

GitHub Copilot

Claude AI

PostgreSQL

Spec-kit

Job description

Java | Fullstack | Front end | Python | Mendix | Oracle | .Net | SAP

The CloudDM project is an AI driven software development aiming to replace the current DM application. We are replicating (feature parity) this 20 year old mainframe application of which we only have documentation, the strategy configurations and a lot of process knowledge with our SME’s.

What we need is an experienced full stack developer of enterprise grade applications. Who has extensive hands-on experience with AI driven software development, preferably with Spec-kit or intent based development. We use both Github copilot with Claude code and AWS Kiro (next to Python, Vue.JS, Postgress, Claude agents and Cloud native development.

Project Background: Debt Management Rebuild

• KPN runs two legacy mainframe applications handling nearly all company revenue
  • Payment management: invoice-to-bank instructions, payment collection, allocation
  • Debt management: strategies to recover unpaid invoices (reminders, subscription limits, etc.)
• Both must be off the mainframe by end of 2026, creating extreme time pressure
  • Late delivery is very expensive; errors could block auditor sign-off on annual report
• Debt management app is being rebuilt from scratch (no access to original code)
  • Manoj is lead architect, using 25+ AI agents for development
  • AI-intent-based, “no vibe coding” approach: no manual code writing

Role Requirements

• Full-stack understanding required (backend, frontend, infra, DevOps)
  • Candidate does not write code manually; must be able to direct and instruct AI agents
  • Must understand what type of code AI is producing and set guardrails/standards
• AWS and Kubernetes expertise required
  • AWS/Kubernetes certifications preferred but not mandatory
  • Must have handled production workloads (at least medium-sized projects)
• Security awareness is a nice-to-have, not mandatory
  • Must be able to resolve findings independently given tight timeline
• No specific programming language required; no manual code review expected
  • Architecture overview and system-level thinking are critical

Reporting directly to Manoj

Stakeholders and Organizational Context

• Collections department is the primary business stakeholder
  • Includes administrative collections staff and agents handling payment arrangements
• Edwin and Manoj jointly responsible for the project

Timeline and Logistics

• Assignment runs to end of February 2026, with possible extension depending on status
  • Final two months (Jan/Feb) reserved for fixes, optimization, and hypercare post-go-live

Desired start date: as soon as possible

• Office days: Tuesday and Thursday in Rotterdam (negotiable)
• Fully English-speaking candidates accepted; Dutch not required

KO-criteria.