Hire a RAG and retrieval engineer on contract
Most wrong RAG answers are retrieval failures. I find where your assistant or agent loses the right context, fix it, and leave you with the evaluation to prove it.
A good fit if…
- Your RAG assistant answers confidently, from the wrong document.
- You switched to a bigger LLM and quality barely moved.
- Your agent calls the right tools with the wrong context.
- You can't say whether the last change to prompts, chunking or models helped.
Look elsewhere if…
- You need a chatbot UI built. I work on retrieval and evaluation, not front ends.
- You haven't chosen a use case yet. Start with a product decision; retrieval work needs real questions.
Deliverables your team keeps.
From brief to handover.
Collect real questions and failure cases. Map your pipeline from ingestion to answer.
Build the test set and baseline. Trace each failing question to the stage that lost the answer.
Fix retrieval in order of impact: hybrid search, query understanding, reranking, then chunking and embeddings.
An evaluation harness in your repository and documentation your team can keep using.
Works with what you run.
What I've built before.
How it's priced.
Not sure where to start? Most teams begin with a fixed-scope search relevance audit.
Starting prices; the final quote depends on scope and follows a free intro call.
Remote, working with teams in the US, UK and Europe. Async-first, with live time during your Pacific mornings, or your morning and early afternoon if you're in Europe. Contracted and invoiced in USD through Infersense Labs LLC (Wyoming, USA).
Questions buyers ask.
Do you work on AI agents, or just RAG?
Both. Agents depend on the same retrieval quality as RAG, plus tool use. At Shopify I worked across the stack behind Sidekick, its AI assistant for merchants, mainly fine-tuning the LLMs that power it.
Should we fine-tune instead?
Usually not first. If the right documents never reach the model, fine-tuning bakes the problem in. I measure retrieval first and recommend training only when the evaluation shows a gap it can close.
Which vector database should we use?
It matters less than the retrieval design around it. I usually work with what you already run.
How long until we see improvement?
The diagnosis in the first weeks usually points to fixes that move the metrics quickly, often hybrid search or query understanding, before any model work.
Tell me what you're hiring for.
A free 30-minute call with me. You'll leave with a clear next step, and an honest answer on whether I'm the right person for it.