Team

The Team

Omri Dahan

Co-founder, CEO

Core part of building five businesses across five industries. Led the commercial charge at Marqeta from the first dollar of revenue until filing for its IPO.

Anthony Scodary

Co-founder, Co-Head of Engineering

Former engineer at NASA JPL, working on missions to Mars and Jupiter. Previously Physics and Aerospace at Stanford. Expert in scaled deep learning for enterprise speech technology.

Nico Benitez

Co-founder, Co-Head of Engineering

Previously at OnLive, a pioneering cloud gaming company. Expert at highly-optimized streaming media, distributed computing, and NLP. Previously CS at Stanford.

Mike Piechna

Co-founder, CRO

15+ years leading revenue and GTM strategy for early-stage companies. Specializes in building scalable growth engines and has successfully led two businesses through acquisition.

Why Guava?

A backronym before it was a fruit. Each letter is the founding principle the team built around.

Generative

Real-time generative voice. ASR, language, and TTS as one integrated system — not three stitched APIs with three latency budgets and three failure modes. The model speaks; it doesn't translate.

Universal

Voice that works on any carrier. Most voice AI sits on top of 1970s narrowband telephony. Guava processes calls at wideband fidelity end-to-end on supported carriers, so the model hears the consonants and sibilants legacy telephony filters out.

16 kHz wideband

Agentic

Tenacity built in. When one path to completing a task doesn't work, the agent tries another. Re-asks. Escalates. Falls back to a human with full context. Voice AI that gives up at the first hard turn is why most deployments fail in production.

Voice

Not a chatbot bolted to a phone line. Voice is the input, the output, the interface, the data — built around how people actually talk to call centers, doctors' offices, claims adjusters, collectors.

API

Built for engineers, not procurement decks. Your EHR, your CRM, your collections platform stays the source of truth. Guava is the voice your stack didn't have.

What we believe.

Production over papers

We optimize for shipping, not publishing. Every model, every feature is built to work under real load.

Ownership over abstraction

We own the full stack — ASR, TTS, language models — because stitching together third-party APIs doesn't scale.

Reliability over novelty

Our customers can't afford downtime. We choose boring infrastructure when it means 99.99% uptime.

Want to join us?

We're hiring engineers who want to build voice infrastructure that actually works.

See Open Roles