Research Engineer, ML Systems & GPU Optimization
Posted Updated
Description
Neuralk is a deep-tech startup building the next generation of foundation models for data science. As a Research Engineer, you'll own the pretraining stack, push inference to production-grade latency, and write low-level GPU code that enables large-scale experimentation.
You will collaborate with the research team to prototype architectures for structured data, develop internal tooling, and help shape a scalable ML infrastructure in a fast-moving environment with high ownership.
Transformer inference optimization: reduce latency and improve scalability for production models. Pretraining optimization: build and maintain scalable training infrastructure (distributed training, mixed precision, gradient checkpointing). CUDA kernel development: write low-level GPU kernels to optimize bottlenecks and data preprocessing pipelines. Architecture experimentation: collaborate with research to implement/train/evaluate novel architectures for structured data at scale. Tooling & infrastructure: develop internal tooling (experiment tracking, evaluation pipelines, benchmarking). Research contribution: publish and engage with the ML research community. Master's or PhD in Computer Science, Machine Learning, or a related field. Equity (BSPCE) State-of-the-art ML infrastructure French level paid leave Offices in Paris and LondonSkills
Sign in to read and join the discussion.