Looking for a job

Nikita Vivat

@nikitavivat

AI Systems Engineer at Expert Invest building multi-agent support systems and cutting routing latency to 6 seconds.

At Expert Invest, I designed key components of AMA Concierge’s multi-agent support platform, including dialogue state management, knowledge navigation, and routing across AI agents, CRM workflows, and human operators. I reworked business-process routing with an SVM classifier, reducing average latency to about six seconds while maintaining about 97% accuracy.

At Pathio, I designed the architecture and built the backend for an AI-powered travel planning platform, including REST APIs, personalization services, and itinerary optimization. Across my work, I’ve delivered production AI and backend systems, from RAG and tool-using agents to deterministic decision engines and integrations.

Experience

Work history, roles, and key accomplishments

Current

AI Systems Engineer

Current

Expert Invest

Mar 2026 - Present (7 months)

Designed and built key components of a multi-agent customer support platform processing ~8.8K messages/day, including dialogue state management, RAG-based knowledge navigation, and routing. Re-architected business-process routing with SVM, reducing latency from 40-80 seconds to ~6 seconds while maintaining ~97% accuracy.

LangGraph RAG Kafka Redis SVM Gemini Embeddings Python fastAPI Docker Kubernetes

Current

AI Systems Engineer

Current

Pathio

Sep 2023 - Present (3 years 1 month)

Designed product architecture and built backend platform including 50+ REST APIs, AI concierge, data-processing pipelines, and personalization services. Implemented personalized itinerary optimization engine and embeddings-based recommendation system.

Python fastAPI REST PostgreSQL Redis Docker Kubernetes LLM Embeddings

AI Systems Engineer

Kommo / amoCRM Global

Sep 2025 - Mar 2026 (6 months)

Proposed and designed architecture for securely executing LLM-generated Python code as foundation of an analytical AI agent. Designed multi-layer sandbox using AST validation, import allowlists, nsjail isolation, and resource limits. Contributed to multi-agent Copilot architecture based on A2A and MCP.

Python LLM Sandboxing AST Nsjail A2A MCP DataFrames Multi Agent Systems

AI Systems Engineer

AI2BIZ

Aug 2024 - Aug 2025 (1 year)

Designed and implemented a multi-channel AI sales funnel with shared customer context, lead scoring, and CRM integrations, increasing qualified leads by 4-5x. Contributed to 10+ production AI solutions including RAG and Tool Calling agents for sales conversations and knowledge retrieval.

RAG Tool calling CRM Lead Scoring Python fastAPI Lang Chain

AI Engineer

Independent

Mar 2022 - Jul 2024 (2 years 4 months)

Delivered 40+ commercial AI and backend projects for small and medium-sized businesses, including AI agents, REST APIs, CRM/payment integrations, workflow automation, analytics, recommendation systems, ranking, and regression solutions.