12+ years turning messy internal operations into predictable delivery: process design, automation, and pragmatic AI. I sit between the business and the build, and I stay hands-on.
If any of these sound like your team, that's exactly the work I own.
When onboarding, approvals and reporting still run on spreadsheets, Slack threads and one person's memory, and every new hire quietly makes it worse.
Internal tools · Ops systemsMulti-hour processes turned into minutes. Manual handoffs and double data-entry removed, so the fastest step becomes the one you no longer do.
Process & workflow automationFragmented data pulled into a clean, governed model with role-based access and reporting leadership actually trusts.
Data modeling · GovernancePractical AI inside real operations, where it removes load with human oversight instead of adding a faster channel of chaos.
AI enablement · AutomationThe unglamorous work is where delivery actually happens: clarifying ownership, removing handoffs, and making progress visible before deadlines become emergencies.
I talk to the people doing the work before I touch a tool. The real process lives in the shadow spreadsheet, not the process doc.
I design the data model like it has to survive audits and growth, not one demo. Most "data problems" are ownership problems in disguise.
One workflow, end to end, in weeks, then iterate with real users. Small, used and boring beats big, admired and abandoned.
Anonymised under NDA: names removed, numbers rounded. The pattern is the point.
The "source of truth" was five disconnected spreadsheets that disagreed with each other. I mapped the real entities, defined ownership, and rebuilt it into one connected model with validation, role-based access and executive reporting.
Redesign before automation: I deleted steps that only existed because of an old mistake, killed double data-entry, and let the system assign work instead of a human guessing. Then automated what was left.
Ran an HRIS selection across a dozen vendors end to end: market scan, shortlisting, technical due diligence and commercial negotiation. Then kept the high-value, non-standard workflows in-house.
My current edge: adding AI to internal operations where it genuinely removes load, like request triage, status summarization and retrieval over internal docs, always with a measurable human check and never as a faster way to hide uncertainty.
The question I keep asking: where does AI genuinely reduce operational load, and where does it just create a faster version of the same mess?
That's the mess I like to turn into a predictable system. Open to Technical Delivery, Automation and AI Enablement roles, remote.