Senior Machine Learning Research Scientist

At Atomwise, we invented the first deep learning neural networks for structure-based small molecule drug discovery and we are currently deploying it in one of the largest applications of machine learning for life sciences. We work on Alzheimer’s, cancer, diabetes, drug-resistant antibiotics, safe pesticides, among treatments for other diseases. We’ve partnered with 4 of the top-10 US pharma companies, raised over $50M from top VCs, and have 100+ diverse projects currently running.

You should think about joining us if you care about making a difference in treating disease and saving human lives. But also, if you are up to tackling some of the hardest open challenges in deep learning today:

  • Non-stationary, unbalanced, and noisy data: Our training data is seldom i.i.d.; new medicines are unlocked by pushing out into newly-discovered biology. Classes are extremely unbalanced, ratios of 1 positive to 70,000 negatives are typical. Help us reason about how to learn appropriately without dismissing nor overfitting to the data; identify when we can trust a label or have confidence in a prediction; and develop techniques to find and correct for systemic biases.
  • Extreme scaling: Medicinal chemists can synthesize about a trillion trillion molecules today. Help us scale predictive algorithms to orders of magnitude beyond those contemplated in any other problem domain today.
  • Multi-parameter optimization: Medicine has to be both safe and effective, so we have to concurrently optimize a number of criteria such as potency, selectivity, solubility, toxicity, synthesizability, etc. Help us efficiently explore the Pareto frontier and avoid mode collapse.
  • Adversarial generation of synthetic data: Data augmentation has shown utility in improving the robustness of predictions. Help us find ways to best integrate molecular physics simulation and machine learning to impute new data.
  • Explainability and visualization: Subtle patterns govern molecular recognition. Help us to understand how they lead to the discovery of fundamental chemistry by AI.

Our Machine Learning team is small and growing quickly. As a result, there are plenty of opportunities to have a big impact on our success.

Required qualifications:

  • Ph.D. or M.Sc. in computer science, statistics, data science, or related field
  • 5+ years of extensive practical experience and proven track record of developing, implementing, debugging, and extending machine learning algorithm
  • Knowledge of modern neural network frameworks such as Tensorflow, Torch, or Theano
  • Strong analytical and statistical skills
  • Scientific rigor, healthy skepticism, and detail-orientation in running and analyzing experiments
  • Familiarity with processing large data sets in a Linux environment

Preferred qualifications:

  • Software engineering skills and coding experience in at least one high-level programming language (Python, R, Java, C/C++, etc.)
  • Biomedical knowledge or experience in processing chemical or biological data is preferred but not required
  • Experience with cloud computing environments (AWS/Azure/GCE)

Please apply with a resume and cover letter.

Compensation and Benefits:

  • Great, world-class team of colleagues – scientists from a variety of backgrounds (chemistry, medicine, biology, physics, CS/ML)
  • Stock compensation plan – you’ll be an Atomwise co-owner
  • Platinum health, dental, and vision benefits
  • 401k with 4% match
  • Flexible work schedule
  • Generous parental leave

Atomwise is not currently offering visa sponsorships for any position. Please only apply if eligible to work in the U.S.

Apply For Job