Endpoint Researcher Lead

About the role

PointFive is building TokenShift, an endpoint runtime that gives enterprises control over AI coding agents on developer machines. As the Endpoint Researcher Lead you will push the frontier of secure sandboxing, local model deployment, and OS‑native isolation across heterogeneous endpoints.

Key responsibilities

• Research and reverse‑engineer AI coding agent runtimes (Claude Code, Cursor, Copilot CLI, Codex) to understand their OS interactions.
• Design and implement sandbox architectures that safely constrain filesystem, network, process, and credential access on macOS, Windows, and Linux.
• Build endpoint harnesses for TokenShift to intercept, control, and modify agent execution flows.
• Develop infrastructure for deploying and managing local LLMs across diverse hardware (CPU, GPU, Apple Silicon) and operating systems.
• Evaluate and benchmark local inference runtimes such as llama.cpp, MLX, ONNX Runtime, and Ollama.

Required profile

• Deep understanding of operating system fundamentals (processes, permissions, filesystems, IPC, networking).
• Hands‑on experience with sandboxing and workload isolation on Linux, macOS, and Windows.
• Investigative mindset with strong reverse‑engineering skills.
• Ability to move between research prototypes and production‑grade code.

Required skills

• Go, TypeScript
• Linux, macOS, Windows system internals
• Container runtimes, Linux namespaces, seccomp, cgroups, AppArmor, SELinux, eBPF
• macOS sandboxing, Endpoint Security Framework, launchd, XPC
• NVIDIA CUDA, Apple Metal, DirectML, ROCm
• Local LLM frameworks: llama.cpp, MLX, Ollama, ONNX Runtime
• Experience with AI coding agents (Cursor, Claude Code, Copilot CLI, Codex)