Egor Pakhomov
Principal Research Scientist
My Background
Egor Pakhomov is a Principal Research Scientist at Salesforce AI Research, where he takes ideas from open question to shipped product. He pre-trained large language models, including Salesforce’s xGen-small family. His recent work spans AI agents: how they remember, how they carry work across days, and how well they use the hardware they’re given. That work, published with open benchmarks and code, shapes how Salesforce measures and improves its own agents, including Agentforce Memory. Before Salesforce, he worked at OpenAI, Airbnb, and Apple.
Egor's latest articles
Blog
Salesforce AI Research doubled Agentforce memory performance by creating a framework that tracks what an AI agent remembers. Imagine the following scenario: In March, a customer tells an agent that their office is…
Blog
Breaking the Memory Trilemma: How to Build AI Agents That Actually Remember
Imagine an AI assistant that forgets your project requirements between Monday and Wednesday, or one that takes 30 seconds to recall a simple preference you mentioned yesterday. This is the reality of AI…
4 authors
Blog
xGen-small: Enterprise-ready Small Language Models
xGen-small is an enterprise-ready compact LM that combines domain-focused data-curation, scalable pre-training, length-extension, instruction fine-tuning, and reinforcement-learning to deliver Enterprise AI with long-context performance at predictable, low cost.
9 authors

