Mark Holland · Senior AI Solutions Engineer · Remote, Florida
I build software people can rely on, and I use AI only where it earns its place.
I've shipped AI that runs every day at a national healthcare data company, and built platforms for law firms, data pipeline teams, and contractors. For smaller businesses, I'll tell you plainly when a simple automation, a custom app, or a better website will do the job.
Watch my work in under 90 secondsOpen to remote AI roles and select projects.
Use Case Platform
Featured work
Three builds that show the range, from enterprise healthcare to a small contractor. Real screens, demo data.
Retrieval · In production
RFP Assist
3 to 4 months → 2 to 3 weeks
Drafts each RFP answer from the approved answer library, shows where the wording came from, and leaves every decision to a reviewer.
My part. I designed and built it and was responsible for testing and release.
Read the case: RFP Assist
Test harness · In production
Application Testing Engine
Every pull request
Runs on its own whenever an app lands in Azure DevOps. It uses the app the way a person would and tells the team what it could not check as well as what failed.
My part. I designed and built it and was responsible for testing and release.
Read the case: Application Testing Engine
Released
Contractor PM Platform
Spreadsheet to one app
A specialty contractor ran its estimating out of a spreadsheet workbook. Now one web app on a laptop, tablet, or phone carries a job from the first call to the final cost.
My part. I built it for them, from the first call to release.
Read the case: Contractor PM PlatformMore in production at work
Built at work for a national healthcare data company.
Intake · In production
AI Use Case Submission Platform
Intake for AI tool ideas. Each one is screened for patient data first, and the ideas it clears are scored against a weighted rubric, with a reason for every rating, and ranked before the review meeting.
See the app: AI Use Case Submission Platform
Signal agents · In production
Payer Churn Signal App
Combines Gainsight, Salesforce, and ServiceNow data with AI agents that scan the news for M&A signals, then alerts account teams to at-risk clients.
See the app: Payer Churn Signal App
Library · In production
Prompt, Skill & Agent Library
A company-wide library of prompts, Skills, skill chains, and agents, with peer voting to keep the most helpful.
See the app: Prompt, Skill & Agent LibraryIn development and prototypes
Each shown at its real stage.
Telemetry · In development
Observability & Governance Engine
An engine that sits between an organization's AI applications and the model providers. It records every AI call with the exact data versions behind it.
See the app: Observability & Governance Engine
Prototype
Stress Tester
A load testing tool for internal applications, built in one session as an MVP for another development team, with real and demo modes, pluggable auth, and full handoff docs.
See the app: Stress Tester
In development
Enterprise Platform MCP
Gives AI assistants safe, controlled access to a large healthcare data platform through one shared core, four adapters, and 13 tools, with a person approving before anything is promoted to production.
See the app: Enterprise Platform MCPIndependent work
Built outside my day job, including one platform built for a client.
PII gate · Launching soon
LawLynx
A multi-agent legal platform with 51 specialized AI agents and a PII gate in front of every external model.
See the app: LawLynx
Model router · Early access
Command Center OS
My own AI operations platform. It runs several AI coding tools side by side, sends each job to the cheapest engine that can do it, keeps a shared memory, and waits for my approval before anything goes out.
See the app: Command Center OS
Audit ledger · Released
PresidioFlow
Repairs records that regulated data pipelines reject, and records every fix in a hash-chained audit ledger.
See the app: PresidioFlow
Personal tool
Cloud CLI Command Forge
A Next.js web app I built that turns plain-English requests into AWS, Azure, and Google Cloud CLI commands and Terraform, using an AI model. A teammate and I used it to cut weeks of Azure work down to a couple of days. It stayed private, shared through a temporary preview link, and was never published.
See the app: Cloud CLI Command ForgeHow I keep AI honest
Five layers between a question and an answer. If one misses a problem, the next can still catch it. And if AI isn't the right tool, I don't use it.
Read how I keep AI honest- Ground itAnswers come from approved sources only
- Cite itEvery answer shows where it came from
- Check itAutomatic checks and a second opinion
- Approve itA person signs off on high-stakes work
- Watch itTest every change, monitor what runs
Writing
- Series 1From AI Idea to ProductionAll 20 parts in six stages, with the full articles and free downloads that go with them.
- Series 2From Automation Idea to Measurable ROIA 20-part series on picking the right work to automate, measuring the baseline before you build, and proving the return in numbers a CFO will accept.
Posts on how agents should behave
- Agent permissions: your AI agent should never approve its own actions
- Durable agent runs
- GraphRAG provenance
- Agent memory
- Exception math
Latest articles
Free downloads
Eval Set Starter KitRun ten example cases against a model, check every answer, and fail the build when quality drops below a frozen baseline.
Claude Code Starter KitStart a project with rules, a deny list, tested hooks, two skills, a verifier, and a tool connection (MCP) setup.
AI Release Pipeline KitShip AI changes only after tests, an eval gate (questions the AI must answer well), scans, and a person's approval.
Hiring?
Senior AI Solutions Engineer, Applied AI, or Forward Deployed roles, remote. My resume is two pages and matches this site.
Have a project?
Not every problem needs AI. Tell me what's slowing your team down, and I'll tell you what would actually fix it, whether that's AI, a simple automation, or something else. You'll get a clear scope and price before any work starts.