Founder

I build products.
I bring people together.

I’m Shubham Joshi, founder of Tacet. I’ve built AI products, started businesses in Japan, and helped create a community of builders at Johns Hopkins. I’m interested in how people will create and make decisions as AI takes on more work.

Start a conversation ↗

What I’m building now

Meet Tacet.

Tacet turns an expert’s decisions into tested rules that can help AI agents and new hires. I founded and built it solo; it is now in early access.

Tacet product demo

What I’ve
learned by doing.

A few experiences that shaped how I approach a problem, a product, and a team.

01

Learn the problem.

I moved from business and analytics into building AI products, learning through research and hands-on work.

02

Make it tangible.

I’ve built an agent-governance prototype, a creative tool, and healthcare software to explore different problems firsthand.

03

Bring people in.

I helped co-found BAAIO at Johns Hopkins. The organization grew to 100+ registered members; I helped lead workshops and a buildathon.

04

Be willing to change course.

I developed a detailed vision for autonomous physical-intelligence labs, then paused it after assessing whether it could be funded. I want to test ambitious ideas against practical constraints.

Earlier experiments in Japan

50 batches.
One lesson.

I tested ice cream recipes with students while exploring a small business in Japan. The product mattered, but licensing and kitchen requirements mattered too. I learned to test those constraints before committing further.

Labeled test batches of ice cream in a freezer
Ice cream experiments in Japan
BAAIO event at Johns HopkinsBAAIO at Johns Hopkins

People and community

Building a
community.

At Johns Hopkins, I helped organize hands-on AI workshops and a buildathon. Participants took ideas from sketches to working prototypes during the event.

Explore BAAIO ↗

Other products
I’ve built.

Two more examples of my product and technical work.

Product · Full-Stack Engineering

AgentDock

A control layer that lets AI agents use tools within defined permissions, budgets, and approval rules.

Built a prototype with approval steps, enforced permissions, and recovery after interrupted jobs.

Live project ↗Demo ↗Source ↗
What I did

Designed and built the application, policy engine, integrations, and job system. It is a prototype; isolated execution and arbitrary external MCP servers remain future work.

AgentDock workspace showing governed agent runs

Product · Creative Technology

Vibe Curator

An editable creative space for combining imagery, music, motion, and effects.

Designed and built a persistent editor with a shared WebGL scene engine.

Live project ↗Demo ↗Source ↗
What I did

Built the concept, interaction design, provider integrations, scene model, and renderer. Generated effects are checked before entering the rendering pipeline.

Vibe Curator promotional preview showing a visual scene and creation prompt

Get in touch

Let’s talk about
what’s next.

I’m interested in meeting investors, fellowship programs, and people who support early-stage founders. I’d be glad to share more about my work and where I’m headed.

Start a conversation ↗