At Blynt, we are building the interface of the future: voice that just works. Our team previously built Sonos Voice Control, delivering a production-grade voice experience to millions of users worldwide with a fraction of the resources of Big Tech. We pushed the boundaries of what's possible, running the entire AI stack directly on Sonos speakers under heavy resource constraints.

Few teams worldwide have shipped a voice experience that combined technical excellence with meaningful adoption. This is the standard we set for ourselves at Blynt - bold science and engineering ambitions within a delightful product experience.

We focus on everything that happens between the microphone and the AI agent: turning raw, continuous audio into structured, real-time events that downstream systems can act on. Our model stack spans speech enhancement, real-time transcription, and turn-taking, with environmental and paralinguistic understanding on the roadmap. Powering the next generation of real-world agentic interactions, from robots to wearables and beyond, requires a fundamental shift in the underlying speech technology. That's what we're here to build.

With more than a decade of experience in Audio and Speech AI, we understand the full stack and we know which techniques work in theory and which survive contact with production. That depth, combined with a product-driven culture and hands-on technical execution, is our edge.

We're based in Annecy and Paris. Remote works for people who thrive in it. Annecy offers a unique environment for those who feel alive when connected with nature and mountains, yet have bold professional ambitions and love in person work.

Key Challenges

• Research: We are pushing the frontier of domain adaptation and contextual biasing for audio and speech models. By combining run-time contextualization and acoustic-domain adaptation, we deliver per-customer, per-session, per-turn personalization, enabling agents to interact accurately and naturally in real-world, noisy, multi-user environments.
• Product: We are building the opposite of a black-box API. Blynt gives control and trust to builders: developers can inspect, debug, and deploy autonomously, identifying and fixing issues themselves. Every decision and every interaction is visible, enabling teams to ship reliable voice experiences at scale.
• Infrastructure: Our systems must run high-performance, low-latency speech pipelines across CPU and GPU, at scale, with full support for adaptive deployment. Reliability, efficiency, and observability are first-class requirements, and building this infrastructure demands deep technical mastery and creativity.

Who we're looking for

We're hiring our first engineers and scientists. You won't be filling a role on an existing team; you'll be building the team, the codebase, and the technical culture from scratch. No established playbook. You'll switch between research prototyping and production shipping in the same week.

You've likely done some of these before:

• Contributed meaningfully to an early-stage technical team where scope was fluid and ownership was real.
• Shipped a real-time AI/ML pipeline to production, and dealt with the messy realities of latency, noise, and scale.
• Built developer-facing infrastructure or tooling where observability and debuggability were core concerns, not afterthoughts.
• Worked with speech or audio ML models (Speech Enhancement, ASR, Keyword Spotting, Neural Audio Codecs, Audio Classification, …) beyond just calling an API.
• Optimized inference or compute-heavy workloads in demanding environments: on-device, or cost-sensitive cloud deployments.

Let’s be clear, we don’t expect anyone to check every box. We don’t really care if you’ve worked specifically on audio before, as long as you’re hyped by our mission and the perspective of building something from the ground up. We care more than anything about who you are, what drives you, and where you’re headed next.