Work

What it looks like in practice

Typical cases for each area of my work. I don't publish real client names — confidentiality first. I'll share details when we discuss your task.

Infrastructure

VPS hardening & monitoring

Task: a server behind a website and mail kept falling over under load, a breach was suspected.

Solution: configuration audit, closing vulnerable services, firewall and brute-force protection setup, monitoring deployment with alerts.

Outcome: stable operation without intervention, problems flagged before users notice them.

AI bots

Telegram AI assistant

Task: an accounting team needed an assistant answering client questions from their regulations.

Solution: an LLM-powered bot connected to the company knowledge base, dialogue history, handing complex questions to a human operator.

Outcome: routine questions closed automatically, operators focus on complex cases.

Automation

Request automation in n8n

Task: a sales team manually copied requests from email and messengers into a CRM.

Solution: n8n scenarios: intake from three channels, AI-based sorting by type, automatic CRM cards and manager notifications.

Outcome: a request reaches a manager in minutes, not hours; nothing gets lost.

RAG

Local knowledge base (RAG)

Task: a law firm needed search across a contract archive without sending data to the cloud.

Solution: a local language model + RAG: documents indexed on the firm's server, answers generated with references to sources.

Outcome: lawyers find contract clauses in minutes, data never leaves the server.

Analytics

Daily digest bot

Task: a product founder wanted a morning overview of trends and mentions without an hour of feed reading.

Solution: a bot collecting data from set sources, AI compressing it into a short summary with key changes, delivered at 8:00 daily.

Outcome: one screen of text instead of a dozen sources — every day, with no human involved.

Infrastructure

Legacy app Docker migration

Task: an old application lived on a single server; updates were a risk and scaling was a pain.

Solution: packing the app and environment into Docker containers, orchestration, automatic backups, migration to a new server with zero downtime.

Outcome: updates stopped being an event, the environment is reproducible, rollback is one command.

Your task may be similar

These cases show the approach, not the limits. Describe your situation — we'll figure out what can be done.