PostEra (YC W20) Is Hiring AI Researchers for Drug Discovery

Role Overview

In this role, you volition create the agentic investigation vertical at PostEra, using agentic systems to automate the betterment of mechanistic models for biochemical and physiological processes, and analyze biologic data for mark validation. You volition engage closely alongside chemists and biologists to use these models to run medication finding decisions. 

You volition additionally create device learning methods that can quickly modify to new medication finding problems from constricted tagged data. A particular concentration is molecular and tabular in-context learning and construction basis models from PostEra’s proprietary multimodal data. You volition build models that use the environment of medication finding effectively, decide which previous examples and tasks are relevant, quantify whenever transfer is helpful or harmful, and provision dependable predictions under allocation shift. 

You volition assistance run the complete investigation loop: defining tasks, constructing datasets and evaluation episodes, evolving powerful baselines, training and scaling models, performing rigorous ablations, and translating prosperous methods into capabilities used by PostEra’s scientists. Prior medication finding cognition is not required, but you have to be motivated to study the domain and activity closely alongside medicinal chemists, computational chemists, and another scientists.

Key Responsibilities

• Agentic Research for Chemistry and Biology: Develop and benchmark agentic systems to automate the betterment of quantitative models to simulate biologic and physiological processes, and the inspection of biologic data. 

• Research Direction and Execution: Independently identify, formulate, and guide investigation projects involving in-context learning, agentic systems, few-shot adaptation, tabular basis models, and molecular device learning.

• In-Context Learning for Molecules: Design and train models that modify to new assays, endpoints, targets, or chemic sequence using constricted tagged environment and heterogeneous historic data.

• Benchmarking: Rigorously difference new approaches against powerful baselines, curating test cases that deconfounds effect of distinct sources of bias. 

• Cross-Functional Collaboration: Work alongside scientists to nexus modeling objectives and evaluation metrics to applicable decisions in potency modeling, ADME prediction, selectivity, guide optimization, and the scheme of first medicinal studies.

• Model Training and Scaling: Develop productive training and data pipelines and, anywhere appropriate, measure models throughout ample collections of molecular and tabular tasks.

• Research Engineering: Produce readable, reproducible investigation code; keep well-tracked experiments; and contribute through code review, documentation, and shared modeling infrastructure.

• Scientific Dissemination: Publish results in foremost device learning, medicinal chemistry, or computational existence discipline venues. You volition be an ambassador of PostEra to the specialized community. 

Key Requirements

• PhD flat in device learning, or STEM investigation involving the betterment of novel device learning approaches 

• Track document of high-quality research, specified as publications and open-source contributions

• Strong investigation or engineering cognition in contemporary device learning, profound learning, or statistical modeling, backed by understanding of the theory rearward device learning algorithms

• Demonstrated ability in at smallest one applicable area: agentic workflows for science, device learning approaches to bioinformatics and medicinal data modelling, in-context learning, tabular learning, few-shot learning

• Hands-on cognition training, debugging, and evaluating ML models in Python using frameworks specified as PyTorch or JAX

• Ability to independently translate ambiguous specialized or specialized problems into well-defined ML projects, including datasets, project definitions, baselines, metrics, and validation schemes

• Ability to scheme careful experiments, benchmarks, and ablations that differentiate improvements from biases, and comprehend which aspects of the example led to the improvements

• Comfort operating in a startup surroundings anywhere priorities evolve, data is imperfect, and high-quality judgement matters as much as raw example complexity

Nice to Haves

Candidates are additional apt to win in this function if they additionally have cognition alongside one or additional of the following:

• Training tabular basis models, particularly for sparse, heterogeneous, small-data, or high-missingness settings

• Developing molecular in-context learning systems or adapting general-purpose in-context models to molecular or specialized data

• Large example training, including 1B+ indicator models, distributed training, sharding, data parallelism, example parallelism, and large-scale data pipelines

• The betterment of AI “co-scientist” systems for bodily or biologic problems 

• Hands-on cognition in modelling biological, biochemical or medicinal data using device learning approaches 

• Moving investigation models into manufacturing specialized application or computational workflows

Salary: 200k - 300k

Equity: 0.05 - 0.1%

Visa Sponsorship: Not at this time

We use recruiting application to assistance the hiring process, specified as collecting, organizing, and managing applications. Applications are reviewed by our recruiting team, and we do not use synthetic intellect or automated decision-making tools to measure candidates or create hiring decisions. Human judgement is used throughout the hiring process, and all final decisions are made by our hiring team. If you would akin additional data concerning how your individual data is processed, or concerning your entitlements under applicable data safety laws, delight communication us.