Senior LLM / AI Engineer
About the role
Location: US and EU engineering hubs (HQ in Atlanta, GA) | Hybrid or Remote | Full-Time | Equity in an early-stage startup | Healthtech
At PromptWrx we're on a mission to fix the financial plumbing of U.S. healthcare. We're not tweaking old-school RCM tools - we're tearing them down and rebuilding smarter with Reve, our AI-driven platform automating every step of the revenue cycle.
We help hospitals get paid for the care they provide, and we take the financial burden off patients - especially when they're not at their best. We do it by outsmarting denial trends, automating appeals, and eliminating rework.
Now we need a Senior LLM / AI Engineer to own the AI behind Reve - the agents that read medical records, interpret payer rules, and decide which denied claims to appeal and how. When they get it wrong, a hospital loses money. We need someone who can design agents that reason well and back it up with evidence. Like the rest of the team, you'll build with AI coding agents (Claude Code and Cursor), which write over 90% of our code.
Why join us?
Competitive pay and stock options - early equity in a growing AI company. Everyone on the team has skin in the game, including you.
You're not babysitting a legacy platform - you'll build at the frontier of AI, healthcare, and revenue intelligence.
Join a leadership team with deep industry and technology experience.
Help solve a $150B+ healthcare problem in a truly novel way.
What you'll do
What you'll do:
Design and build multi-agent LLM systems that analyze claims, apply clinical and payer rules, and write appeal-ready output.
Build retrieval that holds up over large, messy, constantly changing documents - chunking, embeddings, hybrid search, reranking, and citations back to the source.
Stop hallucinations before users see them - grounding, validation layers, confidence scoring, and quality checks.
Own evaluation: golden datasets built with our clinical and coding experts, regression tests on every prompt and model change, and tracking against real appeal outcomes.
Give agents real tools - deterministic services and tool interfaces for anything a model shouldn't guess at.
Handle real-world documents - long medical records, scans, structured healthcare files, and contracts.
Make it production-grade: latency and cost budgets, fallbacks, observability, and a clear trail of why the system made each decision.
Keep PHI safe inside every LLM workflow.
Test new models fast and adopt only what the evals prove is better.
What you'll bring
What we're looking for:
8+ years of engineering, including 3+ years shipping NLP or LLM systems to production.
A strong AI, ML, and NLP foundation. An MSc or PhD in AI, ML, or NLP is a strong plus - shipped work counts just as much.
Production RAG experience - retrieval quality, search and vector stores, hallucination defenses.
Agentic systems experience - multi-step reasoning, tool use, and knowing when a plain workflow beats an agent.
A real evaluation practice - eval datasets, LLM-as-judge calibrated against human review, and regression testing.
Deep hands-on experience with frontier LLM APIs - prompting, structured outputs, long context, and cost control.
Strong Python and the engineering discipline to put models behind clean, reliable services.
You already build with AI coding agents every day and can show us how you direct them and review what they produce.
Healthcare or clinical NLP experience is a plus.
Comfort in startup environments - you move fast, take ownership, make decisions, and don't wait around for permission. Former founders and early-stage startup engineers are a strong plus.
Technologies & Tools:
Claude, GPT, embeddings, rerankers, function calling, MCP.
Search engines and vector stores.
PyTorch, TensorFlow, Hugging Face, OCR and document parsing.
MLOps: Docker, Kubernetes, MLflow, Airflow, DVC, CI/CD.
Our platform: Python, Java, and React on AWS and on-prem.
Required Skills
Required Languages
🇬🇧 English