Sciforium

Pre-training Research Engineer developing and scaling byte-native and multimodal foundation models. The role combines frontier model research with production-grade training code, distributed GPU workloads, and large-scale experimentation.

Build the evaluation, deployment, and MLOps infrastructure that makes multimodal foundation models reliable in production. You’ll enable models on GPUs, automate benchmarking, and connect research experiments to reproducible releases.

Design and optimize high-performance GPU kernels powering large-scale AI training, inference, and real-time applications. This role spans low-level GPU programming, ML framework integration, and end-to-end accelerator performance optimization.

Build and optimize the distributed software infrastructure powering large-scale AI training and inference. This deeply technical role spans CUDA/ROCm runtimes, JAX and PyTorch, multi-node accelerator clusters, profiling, and performance optimization.

Lead the architecture and hands-on development of a high-performance model serving platform for multimodal AI, spanning GPU kernels, distributed inference, scheduling, and real-time APIs. This senior technical leadership role combines deep systems engineering with mentorship and ownership of platform direction.

Build the full-stack developer experience for a high-performance multimodal AI serving platform, including interactive interfaces, developer tooling, billing, documentation, and backend APIs. This hands-on role combines frontend-focused product development with Python services and real-time streaming.

Own the software lifecycle and operational performance of large-scale GPU clusters supporting foundation-model training and inference. This role combines Linux systems engineering, Kubernetes and Slurm orchestration, GPU driver/runtime management, automated provisioning, and distributed-performance optimization.

Build high-performance model-serving systems for real-time multimodal AI applications using C++, Python, and distributed infrastructure. This role combines backend engineering, runtime optimization, and large-scale ML systems work.