Jobs: AI-augmented modelling for healthcare at Pumas-AI

Hi everyone,

I’m happy to announce that the DeepPumas team at Pumas-AI is now looking for two new full-time employees to help revolutionise pharmacology through AI-augmented modelling!

Pumas-AI is a globally remote company that develops software for mathematical modelling within pharmacology and provides expert consulting. Pumas-AI consists of a group of pretty great scientists, many of whom have been deeply embedded in the Julia ecosystem for years. You might, for example, already know of @elrod (LoopVectorization.jl), @andreasnoack, @pkofod (Optim.jl), @mohamed82008 (Turing.jl), @ChrisRackauckas (SciML), Julius Krumbiegel (Makie.jl), just to name a few of the people I have the privilege of working with. We do everything in Julia, and we build on and contribute to large parts of the open Julia ecosystem. Together we have extensive expertise, and we do some pretty cool stuff that ends up having a real impact on people’s health and lives.

In the DeepPumas team, we augment different kinds of pharmacological modelling with ML and SciML to improve model predictivity and our ability to tailor treatment to specific patients. This work comes in three parts. First, we have basic scientific research. The things we do have never been done before, and we need to lead the scientific development of this field. This includes doing research, publishing and presenting at conferences, much like a post-doc would. The second part is software development. New methodology is only helpful if people can actually use it. So we’re democratising the recent advancements through good software, good documentation and good educational resources. And, the third part is hands-on development and application of our science to real-world problems in collaboration with pharma companies and hospitals. This grounds us in reality and keeps us focused on finding solutions that can really be used and have an impact.

The DeepPumas team now has two positions open that I would like to advertise here.

One of the positions is basically to be our ML expert who’s keeping up with all the latest developments in ML and can lead our efforts to squeeze useful information out of different kinds of data. The work would include applying ML to specific data sets to solve particular problems and helping us democratise the process by developing suitable software interfaces and educational resources for DeepPumas users. The data sources will be diverse. For example, in a current project, we’re using interval and rational scale numerical data as well as images. However, that scope is rapidly expanding, and we already have omics and accelerometer data on the horizon, with more to come. As for what kind of machine learning we might use, we’d hope for this ML expert to tell us. However, we already have some interest in CNNs, transformers, graph, and Bayesian neural networks. The problems will be diverse, but they often have a common denominator. We have to figure out how to get something useful from data sets coming from a relatively small number of patients by leveraging prior knowledge about the diseases and treatments we’re studying. The job posting is available here.

The other position is for someone who can lead different DeepPumas consulting projects. At DeepPumas, we are currently taking on few but scientifically challenging and long-running consulting projects. These are projects where traditional pharmacometric modelling has not quite been able to help as much as we would like, where there is data that no one has been able to figure out how to utilise in an effective way, where there’s hope for better individualised healthcare but where the path there is complex and requires a kind of expertise that’s not the primary focus of pharmacology education. As DeepPumas grows and we expand our ability to perform specific analyses quickly, this position might also evolve. Now, however, our projects are similar to post-docs, where we continually update collaborators on our progress but where the project is expected to take more than a year to complete. Every such project needs someone who can take ownership of it. Someone who will be the primary contact to our external partners; who gets to know the biology behind the problem; who charts the course, breaks down the problem into smaller tasks, takes on many of those tasks themselves but outsources the parts that require specific expertise beyond their own. So far, the work is typically done using AI-augmented nonlinear mixed effect models and will thus build on pharmacology, statistics, dynamics and ML. The job posting should be available at in a week or so.

Pumas-AI is a young and growing company. We, employees, all wear many hats and contribute where we can. We are all experts on one thing or another, and we are treated as such. With employees across multiple time zones, the notion of “office hours” has gone out the window and everyone pretty much decides for themselves what hours suit them. We are inclusive and value diversity. Overall, I’d say we’re a pretty fantastic bunch. Get in touch if you’re interested in what we do and would like to work with us.



As part of our effort here, I’ve been working on a new ML library focused on performance for the sort of smaller sized problems we encounter here – often benchmarking at >10x faster than Flux.jl.
We plan to open source it soon (hopefully this week), so there are some exciting developments in this space. :slight_smile:


The job posting for the second position (leading projects) is now up at


Watch Niklas’s video on DeepPumas:

DeepPumas was the start and the foundation of many of the SciML ideas, and is still one of the more active research teams.