Veritone Voice 2.0

Context

Veritone — a US-based AI-for-media company — needed a full product rewrite for Voice. The previous version had UX and architecture limitations that were blocking the roadmap. The challenge: redesign the platform from scratch, with international scale and tight integration into Veritone AI voice pipelines.

My role

Tech lead of the squad: architecture, code governance, reviews, mentoring. I worked directly with Veritone (US) and Fuerza (Brazil) distributed teams via GitLab and Jira, under strict QA processes.

What was built

• Modular architecture in Next.js with decoupled components, ready for multiple white-label clients.
• Audio pipelines integrated with Veritone voice models (synthesis, conversion, custom training).
• Transcription & timeline editor with waveform sync and real-time editing.
• Engineering culture around unit + E2E testing, mandatory code review and CI/CD quality gates.

Outcome

Platform live globally, currently the flagship product line of Veritone's Voice unit. Significant drop in production bugs and a predictable release cycle for the Veritone team.