reasoning Jobs

We are seeking a Senior Scientist to spearhead research in advancing AI models and agents specifically for cybersecurity applications. This role involves enhancing capabilities like security reasoning, code and system understanding, planning, tool use, and long-horizon investigation, with the goal of applying these to critical security challenges. The ideal candidate will drive progress in underlying models, agent architectures, data strategies, training methodologies, and evaluation techniques, while also addressing the unique demands of real-world security environments such as high precision, robustness against adversaries, explainability, and privacy. This position offers a unique opportunity to shape research direction, influence future platforms, and translate foundational AI advancements into practical cybersecurity tools and workflows in collaboration with cross-functional teams.

$224k - $431k

Salesforce AI Research is seeking entrepreneurial Research Interns for Summer 2027 to help build the next generation of Enterprise AI. This is a unique opportunity to work like a founder within Salesforce, exploring high-impact research directions, prototyping cutting-edge AI capabilities, and seeing how ideas are validated with customers and turned into production-ready solutions. Interns will have the freedom to propose and explore new ideas alongside mentors and research scientists, with access to Salesforce's global customer and product ecosystem. The role is ideal for individuals comfortable with ambiguity, bringing energy and focus, and who care as much about customer adoption as technical innovation, offering the pace and ownership of a startup with the reach and resources of Salesforce.

Moveworks is seeking an Agentic AI Harness Architect to define the technical vision for our agentic AI harness. This role involves planning, selecting tools, managing context, and implementing memory, critique, reflection, adaptation, and recovery mechanisms for AI agents. You will invent and prototype algorithms to enable agents to complete complex tasks with minimal supervision and high reliability. The position requires designing model-agnostic abstractions to integrate foundation models, specialized agents, enterprise tools, and workflows, while developing approaches for agent memory and learning that prioritize enterprise security and privacy. You will also create rigorous benchmarks and evaluation methods for agent performance and explore advanced methods like iterative planning, self-critique, and multi-agent coordination. The role includes building prototypes, identifying promising ideas, and collaborating with platform and product teams to bring them to production, while remaining hands-on with code and experiments.

$201k - $352k