Fedor
Pakhurov

LLM systems, agents and applied AI products.

Builds applied AI systems around LLMs, retrieval, agents and multimodal models—with a focus on turning prototypes into useful products.

02 / Selected

Selected areas of practice, from production LLM services to the systems around them.

Applied LLM products

RAG, agents, multimodal prototypes and the production integration required to make them useful.

Multi-agent systems

Supply, transit and export specialists, coordinated around context, proposed actions and explicit confirmation.

Research

Complexity–entropy analysis, Rényi entropy, Bandt–Pompe and clustering for hallucination severity estimation.

03 / Systems

04 / Knowledge

Structured technical collections with prerequisites, equations, permanent URLs and clear progression.

From tokenization and representations to Transformers, RAG and agents.

Likelihoods, latent variables, VAEs, GANs, flows and diffusion models.