The Opportunity

We've fundamentally changed how we build software. Using agentic AI coding systems, our team delivers major platform modernizations and new product modules at a pace that wasn't possible before — not by typing faster, but by thinking clearly: architecting solutions, directing AI coding agents, catching their mistakes, and shipping with confidence.

We need someone who can operate this way independently — driving projects from requirements through production with AI as their primary implementation tool.

What You'll Own

New Product Development (Primary)

• Translate business requirements into technical specs: use cases, data models, API designs, permission matrices
• Design backend architectures following our established patterns — clean layered separation, strong type safety, dependency injection, documented REST APIs — and progressively bring legacy code up to these standards
• Implement by directing AI coding agents — defining the approach, reviewing output critically, iterating until correct
• Collaborate with our Angular Architect on frontend integration

Platform Modernization

• Plan and execute modernization of a mature Python/Flask codebase — evolving it toward clean, layered architecture
• Navigate complex interdependencies — understanding how auth, sessions, permissions, multi-tenancy, jobs, and business logic interact across dozens of modules
• Build comprehensive test suites (regression before, verification during) to make large refactors safe
• Remove dead code, simplify abstractions, and clean up technical debt

Security & Operations Integration

• Proactively identify and address security concerns during development — not just code quality, but real attack vectors and defense-in-depth
• Write operational documentation: changelogs for QA, operations guides for DevOps, architecture docs for developers
• Consider infrastructure implications of your changes: web server configuration, database migration safety, deployment sequencing

Quality & Architecture

• Act as the quality gate for all AI-generated code — catching subtle logic errors, security issues, and architectural violations that look correct on the surface
• Contribute to and evolve our reference architecture and development methodology
• Write documentation that captures why, not just what

Required Skills

Systems Architecture (non-negotiable)

• You can hold a complex system in your head and trace second/third-order effects of changes
• You've decomposed monolithic systems into clean, layered modules in production codebases
• You understand dependency injection, strategy patterns, and when abstractions help vs. hurt
• You have strong opinions about separation of concerns, backed by experience building and evolving real production systems

Python & Web Development (deep, not surface-level)

• 8+ years of production web application experience, with substantial Python work
• Fluent in modern Python: type annotations, generics, Pydantic, dataclasses — you write self-documenting, type-safe code naturally
• Deep understanding of authentication, authorization, session management, and web security
• Experience with relational databases, ORMs, and schema migration strategies
• Understanding of Flask (or similar frameworks), request lifecycle, background job processing (Celery)

Security Mindset (required, not checkbox)

• You've found real security bugs in production code — not just passed a certification
• You understand authentication and authorization deeply enough to reason about attack vectors and defense-in-depth
• You know the difference between "looks secure" and "is secure" — the same skill that's critical for reviewing AI output

Product & Business Thinking (required, not optional)

• You can take a business goal and produce: use cases, workflow diagrams, data models, API designs
• You've acted as a product owner, business analyst, or solution architect — or have the instinct for it
• You understand that technical elegance serves the business need, not the other way around
• You're comfortable with ambiguity and can structure problems before solving them

Technical Leadership

• You've done meaningful code review — not rubber-stamping, but catching real logic and architecture issues
• You take ownership of quality: when something you directed goes wrong, that's on you
• You build tests strategically — knowing what to test, at what level, and when coverage is sufficient
• You can read a 500-line diff and spot the one thing that will break in production

AI-Augmented Development Experience (required)

You have hands-on experience using AI coding tools (Claude, Copilot, Cursor, etc.) for substantial development work. More importantly:

• You understand that directing AI is a leadership skill — you set direction, decompose problems, review critically, and iterate collaboratively
• You treat AI as a capable but fallible collaborator, not a command-line tool
• You brainstorm with AI, ask it to verify your thinking, challenge its suggestions when they feel wrong
• You know that AI output that "looks right and reads authoritatively" can be fundamentally broken — and you have the experience to catch it
• You're excited about the productivity leap, not threatened by it

Culture Fit (this matters as much as technical skill)

• Collaborative over commanding. We've found that people who think with AI — brainstorming, asking questions, building solutions together — are dramatically more productive than people who just issue orders. The same applies to working with humans. We don't want the "smartest person in the room" who dictates. We want the person who makes everyone (and every agent) around them better.
• Ownership mentality. You review work as if your name is on it. Because it is.
• Pragmatic velocity. AI-augmented development means we expect significantly higher throughput than traditional roles — not unreasonable, but meaningfully faster. You optimize for shipping correct code, not for theoretical perfection.
• Intellectual honesty. You say "I don't know" when you don't. You update your mental model when evidence contradicts it. You document trade-offs, not just decisions.

Tech Stack

• Backend: Python 3.11+, Flask, SQLObject ORM, PostgreSQL, Redis, Celery
• Frontend: Angular (handled by dedicated architect — you'll collaborate, not build)
• Infrastructure: Docker, Kubernetes
• AI Tooling: Claude Code (primary development tool)

Who You'll Work With

• Staff Architect — sets high-level architectural direction and cross-cutting design decisions. Reviews your work, available for brainstorming. Your goal is to grow into a peer on your domains, not to depend on daily direction.
• Angular Architect — owns the frontend. You'll collaborate on API contracts and integration, not build UI.
• Business stakeholders — you'll work directly with the business side to understand product requirements, then translate those into specs, use cases, and data models. You're expected to drive this process, not wait for someone to hand you a finished spec.

What You Won't Do

• Build frontend UIs (we have an Angular architect)
• Manage infrastructure or DevOps
• Manage people — you manage AI agents and your own output
• Write code by hand all day — you architect, direct, review, and iterate

What Success Looks Like

At 30 days: You understand the codebase, the domain, and our development methodology. You've paired with the Staff Architect on a task and delivered a meaningful piece of work.

At 90 days: You've independently delivered a significant refactoring project or new feature module. The Staff Architect's reviews find refinements, not fundamental issues. You're contributing to architectural decisions.

At 6 months: You're operating as a peer to the Staff Architect on your domains. You're driving product features from requirements through delivery. The team's throughput has noticeably increased.

This job description was written collaboratively with Claude Code — the same tool you'll use daily in this role.

Engagement model

This is a full time role. Candidates based in Hungary may be employed through AffiniServices Kft. Candidates based outside Hungary may be engaged as full time contractors through Affinitext Inc.