Resume

Igor Fokusov

AI Engineer / AI Solutions Architect
Agentic AI · Enterprise Automation · 1C/ERP

Profile

I design and build AI systems that connect LLMs and autonomous agents with enterprise data, ERP systems and real business workflows.

My primary focus is the engineering layer behind enterprise AI: controlled tool and data access, identity and authorization, agent orchestration, human-in-the-loop workflows, observability, testing and integration with existing information systems.

I have worked professionally in software development since 2004. More than two decades of enterprise automation and 1C experience shape how I approach AI systems: an agent is a participant in an information system and therefore needs explicit permissions, responsibility boundaries, observability and verifiable behavior.

Current focus

  • enterprise AI assistants and agentic systems;
  • multi-step agent orchestration;
  • MCP and controlled access to enterprise systems;
  • identity, authorization, permission propagation and auditability;
  • RAG and enterprise context;
  • AI-assisted software engineering;
  • model routing, independent AI review and evaluation;
  • integration of AI with 1C/ERP and brownfield systems.

Selected AI work

Enterprise AI Assistant — Production

A production enterprise AI assistant with a Telegram interface and 1C integration. Users interact with corporate data and business capabilities in natural language while the agent operates through explicit tools with preserved identity, permissions and auditability. For example, “Who owes us money?” returns a table of all debtors.

Agentic Software Development for 1C / 1c-lite — Working R&D

An agentic development framework for 1C combining specification-driven workflows, independent specification review by a separate model, model routing, parallel implementation, testing, human acceptance and evaluation logging.

Role-Aware AI Assistants for Enterprise Work — Active R&D

Research into AI assistants that account for the user's role, permissions, current task and enterprise state rather than treating each request in isolation. Current directions include AI Accountant and AI Manager with persistent work context, proactive assistance and bounded autonomy.

Professional experience

2025–present · Enterprise AI and agentic systems

Designing AI assistants, controlled agent workflows and integrations that connect models with enterprise systems.

Integration phase · Enterprise automation and architecture

Complex integrations, enterprise automation and architecture for standard and heavily customized information systems.

Foundation · 1C application engineering since 2004

Development, implementation and long-term evolution of 1C applications and business workflows.

Experience across these phases includes:

  • complex integrations between 1C and external systems;
  • HTTP/REST APIs, web services and data exchange;
  • architecture and customization of large enterprise configurations;
  • accounting, management, payroll, production and document-management automation;
  • maintenance and upgrades of heavily customized systems;
  • reporting, business logic and enterprise workflows;
  • access control and enterprise environments;
  • integration of AI into existing enterprise systems.

Enterprise and 1C expertise

1C ecosystem: ERP, Accounting, Payroll/HR, Document Management, Trade Management, Small Business Management, Complex Automation and legacy 1C 7.7–8.3 systems.

Experience includes heavily customized configurations, brownfield systems, managed forms, BSP, data composition, extensions, HTTP services, external APIs, exchanges and enterprise access-control scenarios.

Technology

AI & Agents
LangGraph · MCP · RAG · Qdrant · Langfuse

Backend & Automation
Python · FastAPI · aiogram · PostgreSQL · SQLite · n8n

Enterprise
1C/BSL · HTTP/REST · integrations · Git

AI-assisted Engineering
Codex · OpenCode · model routing · specification-driven development · independent review · automated testing

Education and professional development

Netology — Artificial Intelligence Specialist
Program in progress.

The list of relevant AI/LLM and agent-engineering courses and practical programs is continually updated.

Languages

Russian — native
English — strong technical reading; conversational fluency is actively developing.

Contact