LLM Architect

• Designing cloud-native architectures to run large language models on serverless frameworks (e.g. Kubernetes, Knative, or custom-built FaaS).
• Developing approaches to minimise cold-start latency through advanced container snapshotting, weight pre-loading, and graph partitioning.
• Building distributed inference pipelines with tensor parallelism, model sharding, and efficient memory scheduling to serve LLMs at scale.
• Experimenting with quantisation, pruning, and KV-cache management to squeeze maximum throughput from GPU/accelerator clusters.
• Working closely with applied researchers to turn state-of-the-art methods into robust, production-grade systems.

• Deep understanding of large-scale ML systems engineering, with direct experience in deploying or optimising LLMs.
• Hands-on expertise in C++/Rust/Go for systems programming, plus Python for model integration.
• Strong knowledge of distributed runtimes and scheduling frameworks (e.g. Ray, Dask, MPI, or custom equivalents).
• Experience with GPU cluster management (CUDA, NCCL, Triton Inference Server) and performance tuning across accelerators.
• Solid grasp of cloud-native orchestration (Docker, Kubernetes, Helm) and observability tooling (Prometheus, Grafana, Jaeger).
• Proven ability to translate cutting-edge research into engineered solutions that can scale globally.

• Influence how next-generation LLM services are built and delivered to millions of users worldwide.
• Operate at the intersection of distributed systems, high-performance computing, and AI research.
• Join a global R&D organisation with unmatched resources, where innovation isn t just encouraged it s expected.

Bright Purple is proud to be an equal opportunities employer. We partner with clients who value and actively promote diversity and inclusion across the technology sector.