Research Scientist (Post-Training)

Our goal is an AI for voice conversations that feels like you're talking to a human. We’re building a duplex, speech-to-speech model with full developer control.

MetaVoice is founded by:

* Sid, founding engineer at Wayve.ai ( $1B+ raised)

* Vatsal, co-created Alexa’s first deep learning voice.

We’re passionate about building products people truly love.

Requirements and Experience

* PhD in ML or related field, with publications in top conferences

* 1+ years of experience in training LLM / diffusion / flow-matching on 1B+ param models OR SOTA speech models

* Experience with post-training data curation & supervised fine-tuning of pre-trained models.

* Experience designing & building evaluation workflows – automated or human-in-the-loop.

* [bonus] Early-stage startup experience.

What we offer

* Change the world (when we succeed)

* Environment to do the best work of your life.

* Small team, great people.

* Opportunity to make cutting-edge research work from scratch for production scenarios.

* Work with 100+ TBs of data and 10B+ parameter models.

Culture

* We're an in-person team in San Francisco & love working together. It helps us learn from one another and make decisions quickly.

* We ship fast & obsess over making customers happy.

* We offer high autonomy, allowing everyone to do their best work.

Compensation & Benefits

* Significant equity as a founding team member

* $200K-$500K base

* Immigration support, we are immigrants ourselves

* Fully covered medical, dental, and vision insurance

* 401(k)

Interview Process

We move with speed, and aim to go from first interview to offer within a week.

* Vibe check (30 mins): chat with one of the founders about the vision and your background

* 2 (x45 mins) Technical interviews

* Onsite co-work: Spend half a day with us. We'll build and iterate on a science hypothesis together