Picture yourself here.
LMNT is an early-stage venture-backed AI speech synthesis startup. Our team is obsessed with creating natural, expressive, life-like voice experiences using deep learning as our technology backbone. In previous lives, we have built, productionized, and shipped consumer electronics (e.g. Google Glass), search advertising, and machine virtualization products. We're drawn to hard problems – these seem to be a daily occurrence in our line of work. :)
We’re based in Palo Alto and San Francisco and like working together in-person.
You can see some of our work in our Github repositories.
The role.
As an Applied Scientist, you would research and apply novel techniques in deep learning to build speech synthesis systems.
Responsibilities:
- Contribute to the technical development of LMNT's AI speech synthesis product
- Develop and optimize speech, language, and vision ML models applied to the speech synthesis domain
- Research and develop state-of-the-art deep learning architectures
- Assist in constructing large-scale data pipelines and designing supporting infrastructure
- Contribute to the creation and execution of ML roadmaps in collaboration with the product team
Preferred skills:
- In-depth knowledge of and hands-on experience with autoregressive sequence models and/or diffusion models
- Knowledge of machine learning frameworks (e.g., PyTorch, ONNX)
- Proficiency in Python, C++
- Knowledge of data pipelines and processing for large datasets
- Experience or familiarity with deep learning, especially in audio or speech processing
Signs you’re a great fit:
- Ready to jump into a fast-paced startup; has a sense of urgency to build and ship products
- Paper at top-tier venues (such as NeurIPS, Interspeech, ICASSP, ICML, ICLR)
- Attention to detail in ML model development and performance optimization
- Excellent programming skills, particularly in structural and algorithmic code
- Ability to work effectively in ambiguous situations
- Continuous learner; able to evaluate and acquire new skills/tools as needed
- Willingness to participate in product decisions and explore new directions
Bonus qualifications:
- Love puns, jokes, and solving hard problems
- Knowledge of hardware architectures and instruction sets (e.g., NVIDIA PTX, ARM NEON)
Come work with us!
Send your résumé or LinkedIn URL via email. Suitable candidates will be contacted within a few days.
Note: We value diverse perspectives. If you have a passion for speech synthesis and are willing to learn, we encourage you to apply even if you don’t meet the preferred skills.