Open to FDE roles11 years · Federal healthcare, e-commerce & banking
Forward Deployed Engineer building software with the people who use it.
At Booz Allen Hamilton I work on a VA healthcare claims platform, and I spend much of my time with the people who use it: finding out what they need, building it, showing them, and adjusting based on their feedback.
Integrated health-provider claim-status APIs, built a claims-tracking framework and Tableau monitoring, and shipped React features on a United States Department of Veterans Affairs healthcare claims platform, working directly with stakeholders and claims processors.
3 user rolesVA staff, providers, and veterans now see live claim status, where none was continuously updated before (user-reported).
A Windows desktop app I built that generates realistic synthetic healthcare referral data on demand, then validates, packages, and uploads it, so developers and QA never need production records.
Under 5 minfor a full development database or up to 10,000 records, replacing hours of data prep each day (user-reported).
An AI assistant that reads claim PDFs with Claude, uses prompt engineering and structured outputs to pull out 17 fields, flags missing or invalid information with fixed rules, and drafts the status reply for a person to review and approve.
17 fieldsextracted from each claim PDF, with every reply held for human approval; tested on 60 synthetic claims.
Claude APIPrompt engineeringStructured outputsLLM evaluation
Test planning and validation for banking applications, from requirements through exploratory testing, browser checks, SQL data checks, and release readiness.
4 browsers · 4 environmentsChrome, Firefox, Safari, and Edge checked, with release checks across development, QA, UAT, and staging.
Most of my work is connecting systems, building small tools, and working closely with the people who use them.
Step 1
Listen first
I talk with the people who will use the software and ask how they work today.
Step 2
Define done
I write down the requirements and what done looks like before I build.
Step 3
Build and demo
I build it, show them, and let their feedback decide what happens next.
Step 4
Support adoption
I keep supporting it after release, from troubleshooting to workflow walkthroughs.
After 11 years in development and quality engineering across federal healthcare, e-commerce, and banking, I trace a problem through the whole system, from what a user sees to the data behind it.
When AI fits the problem, I build it the same way: prompt engineering, retrieval over the documents people already use, a person in review, and evaluation for answer accuracy and hallucinations before anyone relies on it.
Core expertise
Engineering toolkit.
The tools and practices I use day to day.
Customer discovery & delivery
Working with stakeholders from the first requirements conversation through demos and adoption.
Application development
I build interfaces and reusable features around the problem a team is trying to solve.
Data and integration
Connecting workflows, checking API behavior, and digging into the data behind what the application shows.
Delivery and quality
Making changes repeatable and giving teams feedback before release.
Generative AI & agents
LLM features built with the Claude and OpenAI APIs: prompt engineering, RAG with embeddings and vector databases, agentic workflows with MCP, and evaluation for answer accuracy and hallucinations. I also use coding assistants daily, then review and test what they produce.
Internal tools & file workflows
Turning repeated manual work into a desktop tool the team can use and maintain.
Web platforms & deployment
What this portfolio runs on: shared content, accessible components, and static builds for Cloudflare or a GitHub Pages path.
Troubleshooting & team enablement
Tracing issues across interfaces, data, and delivery workflows, then helping the team use the fix.
Hire me as a Forward Deployed Engineer, or start with a two-week AI workflow audit. Either way, I work alongside the people who use the system, connect it to your data, and ship something they rely on.
AI workflow audit
Two weeks. One workflow. A plan you can act on.
Fixed price, quoted after a short intro call. For teams that want AI in a real workflow but don't have the people to scope it yet.
01
Week 1
Map the workflow
I sit with the people who run one workflow, such as intake, claims, documents, or reporting, and trace where the time goes, where the data lives, and which rules apply.
02
Week 2
Test where AI helps
I try AI on synthetic or approved data and show where it earns its place, where it doesn't, and what it would take to run it for real.
03
Handoff
Leave with a plan
You get ranked opportunities, the risks, and a scoped first build, walked through with your team so it doesn't sit in a drawer.
Contact
Get in touch.
If you have an integration problem or a Forward Deployed Engineer role, send me an email.