A job description that asks for "LLM experience" will fill your pipeline with engineers who've used the OpenAI playground. A good LlamaIndex job description attracts people who've debugged retrieval failures and know exactly which index type to reach for and why.
What Role You're Actually Filling
Be specific about the RAG use case: document Q&A, semantic search over a database, multi-step agent workflows, or a combination. Include a concrete success metric. "Achieve 90%+ answer faithfulness on our legal document corpus" tells a strong candidate more than "build AI search."
Describe your data environment honestly. What formats are the source documents in? How frequently does the data change? How much volume are you working with? LlamaIndex developers who've worked with messy, high-volume enterprise data think differently from those who've only indexed clean PDF collections.
Must-Haves vs Nice-to-Haves
List the specific LlamaIndex components they need to have worked with: VectorStoreIndex, query engines, agents, data loaders. Include the vector store you use and the LLM provider. "Shipped a production RAG pipeline handling daily active users" is a meaningful qualifier. "Knows about retrieval-augmented generation" is not.
Separate required from preferred. Knowledge of advanced techniques like HyDE or query rewriting is valuable, but if someone has built reliable basic RAG at scale, they can learn the advanced methods. Don't lose a strong candidate to an overly ambitious requirements list.
How to Apply
Ask candidates to describe the most challenging retrieval quality problem they diagnosed and how they solved it. This separates people who've shipped real RAG systems from those who've only done tutorials.
Tell candidates when they'll hear back. "We review applications within 5 business days" sets expectations and signals an organized process. LlamaIndex developers with options move fast.