Stefan
Bocanegra.
I lead product at an AI startup and build my own apps and gadgets on the side.
Entasis
Chief Product Officer since May 2026
Entasis is an early stage enterprise AI startup. I joined in July 2025 as its founding engineer, built our first product (a live view of how each team's work lines up with company strategy), and gave the demo that landed our first client. In May 2026 I became CPO.
I still run most of the demos, 15 and counting for prospective customers, investors and executive candidates. The leadership team also brings me in on roadmap, budget, hiring and fundraising.
- Joined
- July 2025, as founding engineer
- Demos
- 15+, four companies moved on to follow-up talks and one to a formal proposal
- Site
- entasis.ai ↗
FitnessLM
Free iPhone app, in TestFlight beta
FitnessLM covers workout logging, training programs, a food log and a social feed, and it comes with a coach you'd normally pay a subscription for. The coach runs a local Gemma model on the phone itself, about 35 tokens a second on an iPhone 16 Pro, so your data never leaves the device.
I wanted showing up to feel good. Friends see each other's workouts in a feed, rest timers turn into little games, and if you join halfway through the month the app won't hold the days before you signed up against you.
- Built with
- SwiftUI and SwiftData
- Model
- Gemma, running on the phone
- Speed
- About 35 tokens a second on an iPhone 16 Pro
- Status
- TestFlight beta
StoryBox
A storytelling toy for kids, with no screen
A kid taps a toy on the box, a language model writes a short story about that toy, and a text-to-speech voice reads it out loud. I built everything around the AI, from the C firmware on the ESP32-S3 that runs the NFC reader, microphone, speaker and LED ring to the Python backend that asks the model for each story and turns it into speech. I wrote a good share of that code with AI coding tools too.
- Hardware
- ESP32-S3 with an NFC reader, mic, speaker and LED ring
- Firmware
- C with ESP-IDF
- Backend
- Python and FastAPI, with a language model writing each story and text-to-speech reading it
- Parts
- Under $25
- Demo
- Watch it on YouTube ↗
- Code
- GitHub ↗
It took 3rd place and Audience Choice at SCU's Ciocca Pitchathon. I pitched alone against full teams and came away with $600 in non-dilutive funding.
Watch the demo ↗
Senior design
Sponsored by Applied Signal & Technology
My senior design team is building an antenna that follows you. It looks for where your signal is strongest and turns toward it, so the connection holds up as you walk around the room, which a fixed antenna can't do.
Half my degree is electrical engineering, so I spend plenty of time with an oscilloscope, a logic analyzer and LTSpice. Before this I wrote a Verilog model of an SPI SRAM chip, timed to match its datasheet.
- Sponsor
- Applied Signal & Technology
- Status
- In progress
- Classes
- Embedded Systems, Computer Architecture, DSP, Mechatronics
Other builds
Sigma Chi website Visit ↗
As scholarship chair I built our chapter a study resource site and an alumni network that matches 100+ members with 330+ alumni by industry and skills.
SpotifyBank Not public yet
Reads your Spotify listening history and builds a personality profile out of it.
A bit about me.
I'm finishing two degrees at Santa Clara University, computer science and engineering plus electrical and computer engineering, and I graduate in June 2027. Most days I bounce between product work and TypeScript or Python, with Claude Code and Codex open the whole time.
Away from the desk I lift, cook, play video games and watch a lot of movies. Since December 2025 I've also worked with FatRat, an AI consulting and development firm, where I lead a 5 person research and prototyping team and taught an 8 week course for high schoolers on building software with AI.
If you think I'd be a good fit, reach out. I'm always happy to chat.
The real desk.
I rebuilt all of this in Blender from one photo of my desk, caps included.