Speech Recognition Research Lead
Date Posted
25 Jun, 2022
Work Location
Salary Offered
Not Specified
Job Type
Experience Required
3+ years
Remote Work
Not Allowed
Stock Options
No
Vacancies
1 available
At Ello, our mission is to unlock the limitless potential within all children. We build learner-centered experiences, which put children on a path of lifelong learning, creating the foundation for their future and ours.
Ello’s founders have dedicated their careers to pioneering child development at Stanford, Google, and their own successful startups. We were part of Y Combinator’s W20 batch and have just raised a seed round from leading US and European VCs including Human Capital, Project A Ventures, Reach Capital, Khosla Ventures, Homebrew, WndrCo, and K9 Ventures.
About the Role
As Ello’s Speech Recognition Research Lead, you’ll lead our research engineering efforts in speech perception across a variety of tasks that go far beyond ASR. You should have a strong background in applied mathematics, optimization, machine learning, and software development. We are working on research problems on a product development timeline, and you’ll be leading the charge, so you should be enough of a scientist to be able to tackle unsolved technical problems creatively, and enough of an engineer to produce production-ready C++ code. You’ll lead the way both as an architect and by setting an example as an individual contributor who writes great code.
You should have knowledge of both the latest and greatest end-to-end neural speech recognition approaches and older pipelines, hybrid models, and C++ decoders. You should be quite familiar with post-supervised-learning tools like self-supervisions, self-training, and active learning. Ideally, you’ll also have past experience shrinking a large-scale model for real-world deployments on low-power devices through techniques such as pruning, quantization, and knowledge distillation.
Most importantly, you are excited about the opportunity of making an impact on child development at scale.
Ideal candidates for this position are either coming from a research perspective (a PhD or Masters degree focusing on speech recognition or another related area of AI and a publication record at conferences such as ICASSP, INTERSPEECH, ICML, ICLR, and NeurIPS) or have significant ML industry experience (including substantial expertise with sequence modeling either through work in speech recognition or Natural Language Processing). The perfect candidate has a track record of both research affinity and shipping ML products in resource-constrained environments.
Required Experience
- Research experience in speech recognition (ideally demonstrated by publications at venues such as ICASSP, INTERSPEECH, ICML, ICLR, and NeurIPS)
- Familiarity with hybrid pipelines for speech recognition and non-ASR tasks including HMM-FST models and decoder systems
- Experience with a modern Machine Learning toolkit (Python stack, comfortable with TensorFlow and/or PyTorch)
- Excellent Linear Algebra proficiency
- Production programming experience on a large-scale project in C/C++ or similar
- Very accustomed to agile software engineering processes, git flow, etc.
Desired Experience
- Experience with optimizing neural networks to perform well on low-power mobile platforms (e.g. pruning, distillation, quantization) and using tools such as TF Lite/ONNX
- Experience writing high-performance, parallelized code
- Some familiarity with basic full stack web development workflows and frontend API interaction
We look for candidates who
- Have an urge to make a real impact on child development at scale
- Want to do work they are proud of and to ship products they are proud of
- Value open and honest communication that is grounded in empathy
- Want to build meaningful relationships at work
- Take the initiative and take full ownership of their work
- Balance big-picture perspective with attention to detail and creating experiences that will delight our users
Here are some reasons to join us
- Be part of a rapidly growing tight-knit team with the chance to have a big voice
- Tremendous opportunity to become a leader in the company on the back of demonstrated success
- Do meaningful work supporting child development
- Be part of an exceptional engineering team, doing exceptional engineering work
What You’ll Be Building
Reading is the gateway to learning but unfortunately, many children do not receive the individualized support they need to become confident, capable and engaged readers. This creates a cycle where they find reading frustrating and, by avoiding it, fall further behind. Parents want to take action, but they don’t have the experience or time. Enter Ello, an expert coach that makes reading fun and breaks this cycle.
Our first product is an AI reading coach that listens to a child read from real books. It uses evidence-based reading instruction to help them when they get stuck and behavioral psychology to motivate them to keep going.
Who You’ll Be Joining
We’re a small, collaborative team that takes each other’s voices very seriously. Because our mission is to support all children, it is critical that our team is representative of the wider community; we place significant emphasis on ensuring diversity in our team and building an inclusive culture.
We want to create an environment in which we can all be our best selves, use our strengths to take Ello forward, and develop our own skills. We want to have fun while working hard to do impactful things that we’re proud of – we believe it’s possible to do both.