About the job
Why Choose Achira?
Become part of an elite team of scientists, machine learning researchers, and engineers dedicated to revolutionizing drug discovery through predictive modeling.
Explore uncharted territories as we investigate innovative model architectures at the intersection of artificial intelligence and chemistry.
Operate on a large scale with extensive computing resources, vast datasets, and ambitious goals.
Take ownership of impactful projects from inception through to deployment on large-scale infrastructures.
Thrive in a fast-paced environment that values rigor, rapid execution, and a sense of ownership.
About the Role
At Achira, we are developing state-of-the-art foundation models to tackle significant challenges in drug discovery simulation and beyond. Our atomistic foundation simulation models (FSMs) serve as world models, incorporating machine learning interaction potentials (MLIPs), neural network potentials (NNPs), and a variety of generative models.
We seek an exceptional individual who thrives at the crossroads of advanced deep learning architectures and high-performance computing. You will play a crucial role in advancing molecular machine learning by creating high-efficiency implementations of sophisticated architectures for molecular densities, graph neural networks (GNNs), and more — pushing the boundaries of current hardware capabilities.
Foundation simulation models present immense potential in material sciences and drug discovery, yet remain largely untapped. At Achira, you’ll have the chance to change that — enabling models that comprehend and simulate the physical world at an atomic level with unprecedented speed and accuracy.
Your Responsibilities
Architect & Integrate: Develop and implement cutting-edge Graph Transformers, GNNs, and other geometric deep learning architectures into production-ready systems.
Optimize Performance: Enhance end-to-end performance — from high-level implementations in PyTorch/JAX to meticulously tuned CUDA kernels — to maximize throughput, minimize memory usage, and optimize GPU computation.
Scale Effectively: Assist in scaling training and inference workloads across thousands, eventually tens of thousands, of GPUs, maximizing FLOPs, saturating caches, and pushing hardware to its limits.
Collaborate Closely: Work in tandem with cross-functional teams to drive innovation and success in our projects.