NeoCognition
Self-learning AI agents with domain expertise
Member of Technical Staff - Research
About the job
- Strong background in machine learning, natural language processing, or artificial intelligence systems, with experience in large language models.
- Deep understanding of one or more of the following areas: agentic system design, including tool use, planning, reasoning, and computer use; large language model post-training, including instruction tuning, reinforcement learning, and reasoning; or data pipeline design and model evaluation.
- Proficiency in Python and familiarity with modern machine learning frameworks such as PyTorch, JAX, or TensorFlow.
- Demonstrated ability to design, execute, and analyze research experiments from idea to implementation.
- Strong communication skills and ability to work collaboratively in a fast-paced, cross-disciplinary environment.
- Lead research initiatives in the areas of large language model reasoning, post-training, and agentic system design.
- Develop new methods to improve the capability, reliability, and safety of autonomous large language model agents in real-world environments.
- Collaborate with software and platform engineers to prototype and productionize research outcomes into product experiences.
- Design and execute experiments, benchmark performance, and analyze model behaviors to identify failure modes and opportunities.
- Stay abreast of emerging work in reasoning, multi-agent systems, reinforcement learning from human feedback, tool use, and large language model fine-tuning, and contribute to publications or open-source efforts where appropriate.
- Help shape the research culture and technical roadmap of the company as an early member of the team.
- Experience working with open-weight models such as Llama, Mistral, or similar, and training infrastructure.
- Publications in top-tier artificial intelligence venues such as NeurIPS, ICLR, ICML, or ACL.
- Prior experience building research prototypes into usable tools or products.
About the company
NeoCognition builds self-learning AI agents that become specialized experts within defined domains, such as professions, organizations, or software systems. The agents continuously learn from their operating environments to form internal world models and adapt their behavior without full re-training. They aim to develop deep domain understanding and context-aware judgment, enabling expert-level performance in micro-worlds. The product strategy centers on delivering AI agents and augmented workflows for business and professional workflows, along with AI governance and organizational reorganization services. The company’s approach differentiates itself from reactive task-execution agents by emphasizing cumulative learning, specialization, and the creation of agents that learn on the job to improve productivity and generate novel solutions within specific domains.
Company Size
11-50
Company Stage
Seed
Total Funding
$40M
Headquarters
Palo Alto, California
Founded
2026
Get referred to NeoCognition
See people who can refer or advise you
Simplify's Take
What believers are saying
- NeoCognition was hiring four roles in Palo Alto in August 2026, signaling active buildout.
- TechCrunch on April 21, 2026 said about 15 employees are mostly PhDs.
- Vista Equity’s participation opens enterprise distribution through its software portfolio.
What critics are saying
- Ashby listed only four open roles on May 5, 2026, exposing a thin execution bench.
- No customers or revenue were disclosed after the April 21, 2026 launch, extending runway risk.
- OpenAI, Anthropic, and Google can commoditize agent learning before NeoCognition ships a product.
What makes NeoCognition unique
- Yu Su, Xiang Deng, and Yu Gu bring agent research behind OpenAI, Anthropic, Google.
- NeoCognition’s April 21, 2026 debut paired a $40 million seed with deep-pocketed enterprise backers.
- Its continual-learning, micro-world approach targets workflows general agents still mishandle.
Help us improve and share your feedback! Did you find this helpful?
Growth & Insights and Company News
6 month growth
↓ -12%1 year growth
↓ -12%2 year growth
↓ -12%NeoCognition, an AI agent laboratory focused on specialised intelligence, has raised $40 million in seed funding, according to filings with the US Securities and Exchange Commission. Founded by AI researchers Yu Su, Xiang Deng and Yu Gu, the Palo Alto-based company develops AI agents that continuously learn to achieve expert-level intelligence. The startup's mission is to expand access to expertise through these specialised systems. The filing was made under Regulation D of the Securities Act of 1933, which allows companies to sell securities without standard registration requirements. No additional details about investors or valuation were disclosed in the SEC filing.
US AI startup NeoCognition exits stealth with $40m seed. NeoCognition, an AI startup spun out of work by Ohio State professor Yu Su, has emerged from stealth with a US$40 million seed round led by Cambium Capital and Walden Catalyst Ventures. Vista Equity Partners also joined the round, with angels including Intel CEO Lip-Bu Tan and Databricks co-founder Ion Stoica. Su founded the company in 2025 to build self-learning AI agents for enterprises, saying current agents complete tasks as intended only about half the time. Food for thought. Implications, context, and why it matters. NeoCognition's 'expert' agents aim to learn jobs like humans do. * NeoCognition says its agents keep learning how a company works, including processes, limits, and routines, to build a specialized "world model of work" for each setting 1. * The company says this focus can make them quicker and more dependable than general-purpose models. It also says the approach can cut costs and improve safety in high-stakes business use 1. * NeoCognition ties its case to the founders' work on Mind2Web, MMMU, and SeeAct, which cover AI benchmarks and agent studies. It also says work from Su's team is used in frontier large language models from OpenAI, Anthropic, and Google 1. Investors are also backing smaller, task-specific AI approaches. * NeoCognition's funding fits a wider move toward smaller AI models built for one job, which investors believe can beat general-purpose systems in some tasks 2. * Fastino says it trains small models for specific tasks on low-end gaming graphics processing units (GPUs). Yoodli, a communication coaching startup, reached a valuation above US$300 million after a US$40 million series B round by focusing its AI on communication training 23. * The shift suggests companies buying AI care more about dependable performance and lower costs when a system is built for one purpose 2. Stay updated on the go with our mobile app. Get latest insights with smoother, more personalized experience through TIA mobile app. How would you feel if you could no longer use Tech in Asia? Share, tag us, and land on our Wall of!
NeoCognition emerged from stealth with a $40M seed round co-led by Cambium Capital and Walden Catalyst Ventures, with participation from Vista Equity Partners and angels including Intel CEO Lip-Bu Tan and Databricks co-founder Ion Stoica.