The runtime engine (RTE) team is building a distributed inference engine that takes compiled neural networks (generated by our compiler) and runs them on an HPC infrastructure. The goal of the RTE is to accelerate the inference as much as possible, lowering latency and increasing throughput. In order to achieve this goal, we must research and implement efficient techniques for scheduling and distributing data and computation, using intra and inter-node optimizations. The product fulfills Zama’s mission by enabling the use of privacy-presenving AI in practice due to its efficient distribution of heavy computations, making the cost and latency viable for end users.

Your team (and thus you) will be responsible for:

• research and implement ideas to optimize the distribution of heavy computations
• research and implement parallel routines for CPU and GPU
• profiling and optimizing the architecture of the system
• testing the system under adverse scenarios and increase its reliability
• integrating the RTE with the other Zama products

• experience in high performance computing
• experience using C++ or Rust in a production environment
• be based in or willing to relocate to Paris, France
• passionate about privacy and open source

Our process is described in detail here: https://zama.ai/2020/04/28/how-we-hire-at-zama/