Source: Paraform - https://www.paraform.com/share/neocognition/cmpejiwh000020cjykwkd1wy3 Compensation: $280,000-$350,000 base + competitive equity (likely 0.05%-0.5%). Location: Palo Alto, CA - on-site 5 days/week; relocation supported; fully remote only for truly exceptional candidates. Visa: open to transfers and most visas; sponsorship for strong candidates willing to relocate. About this role As a Member of Technical Staff [Research] at NeoCognition, you'll be part of the core team advancing the frontier of LLM agents: systems that can reason, plan, and act reliably in the real world. You'll lead end-to-end research projects, from conceptualization and experimentation to building and testing product prototypes, working closely with engineers and designers. Responsibilities • Lead research initiatives in LLM reasoning, post-training, and agentic system design • Develop new methods to improve capability, reliability, and safety of autonomous LLM agents in real-world environments • Collaborate with software and platform engineers to prototype and productionize research outcomes into sticky product experiences • Design and execute experiments, benchmark performance, and analyze model behaviors to identify failure modes and opportunities • Stay abreast of reasoning, multi-agent systems, RLHF, tool use and LLM fine-tuning; contribute to publications or open source where appropriate • Help shape the research culture and technical roadmap as an early team member Tech stack Python, PyTorch, JAX, TensorFlow, LLMs, RLHF, Llama, Mistral, Open-weight models, Post-training, Agentic system design, Multi-agent systems, AI coding tools About NeoCognition AI agent lab for specialized intelligence, developing AI agents that continuously learn to reach expert-level intelligence. Founding team's research is used in frontier models from OpenAI, Anthropic, Google and others. 10 employees, founded 2025, $40M total funding, Palo Alto. • 2+ years of LLM research (agents, post-training, or evaluations) with publications • Hands-on with post-training (instruction tuning, RL, reasoning), RLHF and open-weight models • Python; PyTorch, JAX or TensorFlow • On-site 5 days/week in Palo Alto (relocation supported)