We are looking for an AI Research Scientist to help advance the modeling approaches at the core of Absentia's platform.
This is a research role for someone who wants to develop new machine learning methods for difficult scientific problems, not simply apply existing models to biological datasets.
Our research problems span molecular representation, graph learning, transformers, multimodal learning, representation learning, uncertainty, mechanistic reasoning, and biological generalization.
The goal is straightforward but difficult: develop AI systems that can reason about how a drug interacts with human biology well enough to make useful predictions before those outcomes are observed experimentally or clinically.
Own ambitious research problems in predictive toxicology, drug safety, and computational biology from hypothesis through experimental validation.
Develop and evaluate novel deep learning architectures and training methods, including graph neural networks, transformers, multimodal models, generative approaches, and other emerging architectures where scientifically appropriate.
Investigate representations that connect molecular structure, biological targets, pathways, dose and exposure, pharmacology, toxicology, and clinical outcomes.
Develop approaches for learning from heterogeneous, sparse, noisy, and partially observed scientific datasets.
Explore methods for cross-domain and out-of-distribution generalization, including prediction on novel compounds, chemical spaces, biological contexts, and endpoints.
Design rigorous experiments, benchmarks, ablations, and evaluation frameworks that distinguish genuine biological generalization from memorization or dataset artifacts.
Develop methods for uncertainty estimation, calibration, applicability-domain assessment, and confidence-aware prediction.
Investigate approaches for making model predictions more mechanistically interpretable, including identifying biological pathways, targets, systems, and evidence contributing to predicted outcomes.
Qualification
Most importantlyA PhD in machine learningExperience with one or more of
Required
You are a researcher who is comfortable working where the correct modeling approach is not yet known.
You care about understanding why a model works, where it fails, and whether it is actually learning something generalizable.
You are comfortable moving between mathematical ideas, experimental code, scientific literature, and large-scale empirical results. You can pursue a research direction independently, but you also enjoy working closely with engineers and domain scientists to turn research into systems that matter.
Most importantly, you want your research to have consequences beyond a benchmark.
A PhD in machine learning, artificial intelligence, computer science, computational biology, computational chemistry, applied mathematics, statistics, or a closely related field, or equivalent demonstrated research experience.
A strong research track record demonstrated through publications, novel methods, significant open-source research, or technically substantial research projects.
Deep understanding of modern machine learning and deep learning.
Hands-on experience developing and evaluating models in PyTorch, JAX, or equivalent frameworks.
Experience with one or more of: graph neural networks, transformers, representation learning, multimodal learning, generative modeling, self-supervised learning, probabilistic modeling, or foundation models.
Strong experimental instincts and experience designing controlled evaluations, ablations, and reproducible research.
Ability to reason carefully about dataset construction, leakage, confounding, generalization, and evaluation methodology.
Ability to independently identify important research questions and carry projects from initial hypothesis through rigorous experimental results.
