Research Scientist, Speech Recognition

Key Skills

Python Programming, Scala Programming, C++, Java Programming, Machine learning techniques, Data science techniques, SQL, Apache Hadoop, MapReduce, TensorFlow, PyTorch, R Programming

Job Description

When you think about your last doctor’s appointment, how much do you actually recall? If you’re like most people, you’ll likely forget over 40% of the conversation. Details fall through the cracks, generic printouts you receive end up in the trash, and important information never makes it to other doctors. Ultimately, you’re left needing to connect the dots in your care. That’s why we started Abridge with a simple thesis: if we can shift agency to consumers, we can help them stay on top of their health. When you use Abridge to record your health conversations, the conversations themselves become your medical record. It’s as easy as that.

From transcribing medical parts of the conversation to delivering key takeaways, our trailblazing work in machine learning research makes the Abridge experience possible. We're currently looking for a research scientist with experience in speech recognition to help us improve the performance and robustness of our transcription modules.

What You'll Do:

• Research and implement speech recognition algorithms and methods
• Create language models from large corpora of medical text data
• Perform data preprocessing and define performance measures based on development and test sets
• Implement and improve tools, algorithms, and prototypes
• Be responsible for measuring and optimizing the quality of your algorithms

Who You Are:

• Graduate degree (MS or PhD) in Electrical Engineering, Computer Sciences, Mathematics, or similar with specialization in speech recognition or machine learning
• 2+ years experience in building speech recognition systems
• Deep knowledge of ASR technologies such as acoustic modeling, language modeling, neural networks, HMM, WFST, and feature extraction
• Hands-on experience in ASR tool kits such as Sphinx, Kaldi, HTK, or Julius

Ideally, You Have:

• Research work or publications pertaining to applying deep learning methods to speech recognition
• Deep fluency in academic fields relevant to speech recognition

Life at Abridge

At Abridge, we’re driven by our mission to bring understanding and follow-through to every medical conversation. Our culture is founded on doing things the “inverse” way in a legacy system—focusing on patients, instead of the system; focusing on outcomes, instead of billing; and focusing on the end-user experience, instead of a hospital administrator's mandate.