AI Research Lead

About the role

PointFive is building the AI Efficiency OS that optimises AI spend across cloud and on‑premise environments. As an AI Research Lead you will drive the research agenda that evaluates, optimises and deploys large language models for real‑world software‑engineering workloads.

Key responsibilities

• Define and lead the LLM research roadmap covering model evaluation, inference optimisation, local deployment and routing strategies.
• Design rigorous evaluation frameworks and benchmarks that reflect actual developer workflows.
• Research model‑efficiency techniques such as quantisation, distillation, speculative decoding and KV‑cache optimisation.
• Investigate when smaller, locally‑deployed models can replace expensive frontier APIs without degrading task quality.
• Develop model‑routing strategies that balance latency, cost, privacy and hardware availability.
• Collaborate with engineering and product teams to translate research findings into production features.

Required profile

• Deep understanding of modern LLMs and transformer architectures.
• Hands‑on experience evaluating frontier and open‑weight models.
• Expertise in inference behaviour, batching, KV‑caches and token‑generation performance.
• Proven ability to design controlled experiments and draw conclusions from noisy data.
• Strong Python skills and familiarity with the current ML ecosystem.
• Comfort with ambiguous research problems and bridging research to production.

Required skills

• Python programming
• LLM evaluation and benchmarking
• Model quantisation (FP8, INT8, INT4, AWQ, GPTQ, GGUF)
• Model distillation, LoRA, fine‑tuning
• Inference frameworks: vLLM, SGLang, llama.cpp, MLX, TensorRT‑LLM, Ollama, ONNX Runtime
• GPU performance optimisation (CUDA, Metal, ROCm, DirectML)
• Distributed inference and multi‑GPU serving
• Model routing, mixture‑of‑models and adaptive inference
• Reinforcement learning, DPO, GRPO
• Agentic systems and coding‑agent evaluation