Machine Learning

About WinterLight Labs and the role

WinterLight Labs is a Toronto-based company which has developed software for objectively quantifying speech and language to detect and monitor dementia and other neurological disorders, using cutting-edge natural language processing algorithms and machine learning. Subsequent to two seed rounds, we are piloting our technology with pharmaceutical companies, senior care providers, and clinicians throughout Canada, the United States, and Europe.

We are seeking a Machine Learning engineer to help with the development of automatic speech recognition (ASR) and machine learning (ML) models.

Benefits include: flexible work hours and location, 4 weeks vacation, equity and more.

This role will involve:

  • In-house implementation of state-of-the-art ASR models, using deep neural networks and open-source tools, such as Kaldi.
  • Research and development of methods for online re-training and adaptation of ASR models, especially for individuals with speech disorders (e.g., aphasia) and elderly voices
  • Analysis of the robustness of lexico-syntactic features derived from errorful ASR transcripts.
  • Implementation of additional lexico-syntactic, semantic, and pragmatic features derived from textual transcripts
  • Implementation of additional acoustic features derived from the speech signal
  • Optimization of feature selection and dimensionality reduction techniques
  • Optimization of model hyperparameters, and experimentation with a variety of model types.
  • Optimization of multi-view approaches to data fusion
  • Leveraging unsupervised and supervised approaches to data fusion

What you’ll need:

  • Graduate degree in Computer Science, Computer Engineering, or an equivalent degree.
  • 5+ years research experience in machine learning, especially in computational linguistics.
  • Research experience in, or demonstrable familiarity with, standard ASR architectures.
  • Research experience in, or demonstrable familiarity with, standard feature toolkits in computational linguistics, such as NLTK, CoreNLP, gensim, or Spacy.
  • Research experience in, or demonstrable familiarity with, modern approaches to deep learning.

WinterLight Labs
Liam Kaufman, CEO
liam [at]