For teams sitting on docs, tickets or product data that people keep asking the same questions about.
Conversational assistants that answer only from your own content and can call your internal APIs to actually get things done — not just chat. Deployable on your own infrastructure so nothing leaves your walls.
What you get
A working assistant in your site, app or Slack
An ingestion pipeline that keeps it current as content changes
Source citations on every answer, plus guardrails against hallucination
Hosted models (Claude, OpenAI) or fully local ones (Ollama) — your call
An admin view for prompts, logs and evaluation
Recent example · The assistant on this site: it answers from ~50 articles and case studies, streams its replies and cites sources — built on Cloudflare Workers AI + Vectorize, re-indexed automatically whenever the content changes.
Workflow automation
For operations, finance and support teams losing hours to copy-paste between tools.
I map the repetitive process end to end, then automate it with n8n and custom code — approvals, invoicing, data sync, notifications, reporting — with monitoring so you know when a human is actually needed.
What you get
A documented map of the current process and the automated design
Production n8n workflows, self-hosted or cloud
Integrations with the tools you already use (REST, GraphQL, webhooks, DB)
Error handling, retries and alerting
A runbook so your team can maintain it
Recent example · A voice-to-task pipeline: dictate a note, it is transcribed with Whisper, structured by an LLM and filed as a ClickUp task with the right project and owner — no more manual entry.
Healthcare integration — HL7 & FHIR
For clinics, labs and health-tech vendors that need to exchange data with hospital systems and EHRs.
Interoperability is my core specialty: HL7 v2 and FHIR R4 pipelines, Mirth Connect channels, validated conversions and safe migration between critical systems.
What you get
Mirth Connect channels — or a Spring Boot microservice — for the interface
HL7 v2 ↔ FHIR R4 mapping with conformance validation
Support for ADT, SIU, ORM/ORU and custom message types
A test harness with sample messages and per-record error reporting
Documentation an internal team can operate
Recent example · A FHIR R4 bundle generator that turns internal records into HL7-conformant bundles, plus Mirth channels linking a scheduling system to a hospital referral platform for secure exchange of availability data.
For companies whose data is trapped in a legacy system — FileMaker, an old ERP, spreadsheets — and cannot reach modern tools.
Reliable sync layers between old and new systems: real-time where it matters, batch where it does not, so you get one consistent source of truth and analytics-ready data.
What you get
A bi-directional sync service between your systems
Real-time (CDC / webhooks) or scheduled ETL, whichever fits
Data cleaning, de-duplication and validation rules
A clean REST / GraphQL API over the consolidated data
A cutover plan with rollback safety for migrations
Recent example · A Node.js service keeping a FileMaker database and PostgreSQL in sync in real time, so a modern web/mobile app and the legacy back office never disagree.
Custom software
For teams where off-the-shelf tools do not fit the workflow and a spreadsheet has quietly become the system of record.
Full web applications — customer portals, internal tools, dashboards, APIs — built to be secure, tested and maintainable, and handed over so you are never locked to me.
What you get
A web app: React front-end, Java/Spring Boot or Python/FastAPI back-end
Authentication, roles and an audit trail
Automated tests and a CI/CD pipeline
Dockerised deployment on your cloud or on-prem
Source code, documentation and a hand-off session
Recent example · BurgerBase — a self-hosted backend (auth, dynamic REST API, sandboxed functions, deploy CLI) that replaced a SaaS dependency for several clients’ React apps, running on Windows Server and Linux with FileMaker, SMTP and LLM integration.
Training & advisory
For teams that want to build AI and automation capability in-house, or need an outside read before committing budget.
I have taught at university level and run professional workshops. I also do architecture reviews and honest build-vs-buy assessments — sometimes the right answer is “don’t build this”.
What you get
Hands-on workshops: AI, RAG, systems integration, testing
An architecture review with written findings and priorities
A build-vs-buy assessment with a clear recommendation
An AI roadmap: what to do first, what to wait on
Follow-up support while your team executes
Recent example · Regular internal training for a dev team on applied AI and integration practice, plus recurring architecture guidance as their systems grew.
Tools & platforms I build with
Python
OpenAI API
LangChain
Hugging Face
n8n
AWS
FastAPI
Docker
Azure
PostgreSQL
MongoDB
ChromaDB
Spring Boot
Java
HL7
React
JavaScript
FHIR
Ollama
Mirth Connect
How it works
From first call to hand-off
A predictable path, low-risk at each step.
1
Intro call — 30 min, free
You describe the problem. I tell you honestly whether I am the right fit and roughly what it would take.
2
Discovery — 1–2 weeks, fixed price
I map the workflow, define what success looks like in numbers, and deliver an architecture and plan — yours to keep, with or without me.
3
Build — iterative
Short cycles. You see working software regularly, not a big reveal at the end. Everything is tested and documented as it is built.
4
Hand-off
Source code, documentation and a walkthrough so your team owns it. Ongoing support afterwards only if you want it.
Engagement models
Every engagement starts with a short call to check it's a good fit. Pricing is scoped to the work — ask for a quote.
1–2 weeks
Discovery sprint
A fixed-scope assessment: I map the workflow, define success metrics, and deliver an architecture and plan you can act on — with or without me.
4–12 weeks
Project delivery
End-to-end build of a defined system, delivered iteratively with tested, documented components and a hand-off so your team owns it.
Monthly
Ongoing partner
A recurring block of time for evolving systems, new automations, monitoring and support — sized to what you actually need.
Pricing
How pricing works
The intro call is free. Discovery is a small fixed fee. After that it is either a fixed project quote — when the scope is clear — or a monthly retainer for evolving work. You always know the number before anything starts, and there are no surprise invoices.