Andrew Scott

I am Andrew Scott. I am a senior software engineer with around 9 years of experience in backend and AI work, and the last few years I have been focused on LLM and RAG systems, mostly on AWS.

Here is what I do best:

Build LLM apps end to end: Python services (FastAPI), tool and API integrations, retrieval pipelines (embeddings and vector search), and solid evals before release.

Ship production systems with Docker, Kubernetes, Terraform, and good monitoring so things stay measurable and reliable.

Keep performance and cost in mind, like tracking latency and token costs and tightening context and retrieval when it matters.

A few achievements I am proud of:

At Phamily, I led the build of an LLM + RAG intake and triage assistant (CareFlow Copilot). It cut average triage handle time by about 18% and lifted completed intake forms by about 9%.

At Deloitte, I helped build a document understanding and scoring platform for claims, reducing manual review time by about 22% and lowering inconsistent decisions by around 10%.

At Gusto, I co-led a rebuild of payroll onboarding that increased first month payroll activation by roughly 11% and cut onboarding support tickets by about 8%.

What I am interested in next:

I like roles where I can build practical AI that automates real workflows, integrate data and APIs cleanly, and get it into production with strong testing and observability. I am especially into LLM apps, RAG, automation, and developer tooling (I have used Cursor in my workflow too).