AI Developer II

Grade Level: Level II

Minimum Experience: 5–7 Years Minimum

WHAT THIS ROLE ACTUALLY IS

You are a senior-level AI engineer who builds production systems — not someone who fine-tunes models in a notebook and calls it a day. You take a written technical specification and you execute it to the letter, on time, with zero defects on first submission. You do not ask for clarification on a spec written clearly. You do not submit work you have not personally tested against every acceptance criterion. Every line of code you write runs in a production environment serving enterprise clients or government agencies.

WHAT YOU DO EVERY SINGLE DAY

• Read the written technical specification before writing a single line of code — understand the deployment environment, security controls, performance benchmarks, and every acceptance criterion before you begin
• Build production-grade AI applications in Python and Node.js — APIs, LLM integrations, and data pipelines that run under real load with real users
• Integrate LLMs into production applications — prompt engineering, context management, token optimization, error handling, and fallback logic that handles every failure mode
• Build and maintain RAG pipelines — vector database configuration, embedding generation, retrieval optimization, and response quality evaluation in production
• Personally implement all security controls specified in the task brief — authentication, authorization, audit logging, data encryption, and input validation — before submission. Security is part of your definition of done
• Write unit tests and integration tests for every component — minimum 80% coverage before submission. Untested code is unfinished code
• Submit work for review only when it passes every acceptance criterion — you run every check yourself before the Head of AI sees a single line
• Document every system you build — architecture decisions, API contracts, environment variables, deployment steps, and known limitations. If you are unavailable tomorrow, someone else can operate what you built today
• Submit a written daily standup update every working day — what you completed, what you are working on, what is blocking you. Three sentences minimum, every day without exception

TECHNICAL REQUIREMENTS — CORRECTED FOR GRADE LEVEL II (5–7 YEARS MINIMUM)

Core Languages (6 yrs Python / 4 yrs Node.js / 3 yrs React): Minimum 6 years of professional production Python — FastAPI, Flask, or Django. You have built, scaled, and maintained Python-based systems under real production load. Minimum 4 years of Node.js — backend services, REST API design, and asynchronous programming patterns. Minimum 3 years of React for production frontend development — state management, performance optimization, and component architecture. You do not look up syntax. You have written enough of it that it is muscle memory.

AI and LLM Stack (4 yrs AI systems / 2 yrs direct LLM / 2 yrs RAG): Minimum 4 years of hands-on experience integrating AI and machine learning systems into production. Minimum 2 years of direct LLM integration in production — OpenAI, Anthropic, or open-source model equivalents deployed to real users. You have handled prompt engineering, token optimization, context window management, fallback logic, and cost management under real traffic. Minimum 2 years building RAG pipelines with a real document corpus and real query volume. You have debugged a retrieval failure in production. You know exactly what caused it.

Deployment and Infrastructure (4 yrs Docker / 3 yrs CI/CD / 3 yrs Cloud): Minimum 4 years of Docker and containerization — production Dockerfiles, multi-stage builds, compose configurations. Minimum 3 years of CI/CD pipeline ownership — you have built and maintained GitHub Actions or equivalent, configured environment promotion gates and automated test requirements. Minimum 3 years of cloud infrastructure on AWS, GCP, or Azure — deployment architecture, IAM permissions, networking, and production incident management.

Security (5 yrs secure coding / 2 yrs regulated environment): Minimum 5 years of secure coding practice — OAuth 2.0, JWT, RBAC, session management, input validation, secrets management, and dependency vulnerability management are all implemented as standard practice. Minimum 2 years building in government or enterprise compliance contexts — FedRAMP-aware architecture, NIST 800-53, CUI data handling, or SOC 2. You have supported or contributed to an Authority to Operate process.

Government and Enterprise (2 yrs regulated environment): Minimum 2 years of direct experience building AI or software applications for government clients or in government-adjacent regulated industries. You understand what documentation a compliance officer needs and you produce it without being asked.

WHAT WE REQUIRE — NO EXCEPTIONS

• 5–7 years building and shipping AI applications in production — real systems, real users, real uptime requirements across multiple organizations
• You have integrated at least one LLM API into a production application and you can describe the specific architecture, the failure modes you encountered, and exactly how you resolved them
• You have built at least one RAG pipeline with a real document corpus under real query volume — not a tutorial, a production system you maintained and debugged
• Your code is clean, documented, and tested before you submit it — not after review feedback. This has been your standard for at least 4 years
• PST morning overlap — Warsaw 1pm to 5pm minimum, every working day without exception
• Professional written English — daily standups, technical documentation, and task communication are precise and unambiguous
• You work async-first. You communicate proactively in writing. You do not wait to be asked for a status update
• HOW TO APPLY Equal Opportunity Employer
• We hire for precision, technical depth, and character — regardless of nationality, background, or location. If these standards match how you already work, we want to hear from you.
• Submit your CV, a link to your GitHub, and a written paragraph — not bullet points — describing one AI application you built in production. Include what the system does, how it handles failures, and one specific technical decision you would make differently today. Applications without the GitHub link and written paragraph will not be reviewed. Shortlisted candidates complete a 72-hour technical build challenge with a mandatory security requirements component.