I'm interested in the point where business processes and technology meet.
I design automation systems that simplify repetitive work, connect tools, and make workflows easier to operate and measure. Most of what I build sits between systems that were never meant to talk to each other — a form, a CRM, an inbox, a database, a model.
My portfolio is deliberately small and deliberately detailed. Rather than list a large number of projects, I document a few systems fully: the manual process they replace, the rules they encode, the decisions they escalate, and what they do when an integration fails.
I care about the operational side as much as the technical one. An automation that nobody trusts gets switched off, so I spend as much time on checkpoints, reporting and edge cases as on the happy path.
How a project usually runs
- Map process
- Define rules
- Design system
- Build
- Test failures
- Measure
Understand the problem before choosing the tool
The tool is the last decision, not the first. Most of the value shows up while mapping the process, before anything is built.
Automate processes, not chaos
Encoding a messy workflow makes it faster and harder to change. Cleaning up the process first is usually the bigger win.
Keep humans in the loop where judgement matters
Irreversible actions, low-confidence decisions and anything that goes out in someone's name keep an approval step.
Reliability matters more than impressive demos
A workflow is only finished when the retry path, the duplicate case and the API outage are handled.
Orchestration
- Workflow orchestration
- Webhooks
- Schedulers
- Queues
Data
- PostgreSQL
- JSON schemas
- REST APIs
- OAuth
AI
- OpenAI API
- Structured outputs
- Classification
- Extraction
Business tools
- HubSpot
- Email platforms
- Helpdesk APIs
- Analytics
If a process in your business runs on copy-paste, it's probably automatable.
Let's Talk