Voice AI
Maya, a live voice-AI appointment-setting agent
Built and deployed a live outbound voice agent for a client, integrating ElevenLabs, Make.com, Twilio and Google Workspace into one no-code solution that handles the voice conversation, scheduling, follow-up SMS and alerts end to end.
The hard part wasn't the technology. It was designing the conversation so a real person on the other end found it clear and easy — the same problem as teaching someone to use a product.
Teaching & enablement
Turning dense manuals into material people can learn from
Preparing for a professional licensing exam, I used a carefully prompted ChatGPT assistant and NotebookLM to turn hundreds of pages of dense manuals into plain-language summaries and audio people could listen to while at work. Others preparing for the same exam used it to cut their study time.
The useful part wasn't the tools. It was saving working professionals real study time, turning dense reading into audio they could get through during the workday.
First-hand as a user
An AI assistant for my own job search
I built an AI-assisted pipeline that researches employers, scores roles against my own criteria, and drafts application material, always under a review step I control before anything goes out. I built it to solve my own job search, which gave me a first-hand feel for the kind of AI-assisted workflow Jobscan builds for job seekers.
It taught me what it actually feels like to be the user of an AI job-search tool: where it helps, and where it has to stay honest and keep a person in control.
How I work
I build with no-code and AI-assisted tools: ElevenLabs, Make.com, Twilio, Google Workspace, and models like Claude and ChatGPT. I pick and connect them to solve a specific problem, not for their own sake.