Job Description & Details
This is a hands-on contract role focused on building and shipping production-grade LLM and machine learning systems. You will not just be writing experimental scripts; you'll be actively deploying architectures that scale on Azure and handling real-world production constraints.
What You'll Actually Be Doing
Your day-to-day will involve designing RAG pipelines, integrating proprietary or third-party LLMs like OpenAI and Anthropic Claude, and managing containerized deployments via Docker and Kubernetes. You will balance modern AI-assisted coding tools like Cursor and Claude Code with robust software engineering fundamentals to ensure code quality through rigorous Git workflows and CI/CD pipelines.
The Core Tech Stack
Python is your primary language here, backed by core ML libraries like PyTorch, scikit-learn, and pandas. You need solid experience with vector stores for RAG implementations and deep familiarity with Azure infrastructure such as AKS, App Service, and Azure AI services. If you haven't shipped production LLM systems or lack containerization experience, you will struggle with the infrastructure demands of this role.
Interview Expectations
Expect to be grilled on your RAG architecture decisions, specifically how you handle chunking strategies, embedding models, and vector database retrieval latency in production. The interviewer is secretly trying to figure out if you actually know how to debug hallucination and latency issues, or if you only know how to run basic API wrappers. Be prepared to explain trade-offs between managed AI APIs and self-hosted open-source models.
Application Advice
Your resume needs to clearly show your hands-on experience deploying AI systems, not just academic projects. Make sure keywords like 'Azure', 'PyTorch', 'RAG', 'Docker', and 'CI/CD' are front and center to pass the automated filters. Emphasize your production track record and your ability to communicate complex technical trade-offs to non-technical stakeholders, as client-facing skills are heavily weighted here.