I am an engineer who turns "I want this to run by itself" into a working product. I design AI agents that think, search, call APIs and see a task through to the end. I write code in twelve languages — from Python and TypeScript to Solidity and C++. I build everything: from a landing page over a weekend to a distributed system with queues, a vector database and monitoring. If a task can be automated, I will automate it. If it cannot, I will find a way.
Senior developer and AI engineer. I started with classic backend work and 1C, moved into machine learning, and settled somewhere between LLM agents and blockchain. Today I take projects end to end: architecture → code → deployment → support.
I enjoy projects where I have to learn someone else's domain and then make it so a person presses one button and everything happens. Legacy code, research and deadlines do not scare me.
Areas I cover end to end. A project is usually a mix of several.
Agents that actually work: they plan steps, call your APIs, read databases, write to the CRM, make decisions and report back. Function calling, tool use, memory, RAG over your documents, autonomous pipelines and a human in the loop where it matters.
I remove manual routine: connecting CRM, spreadsheets, messengers, email, warehouses and payment systems into a single flow. n8n, Make, custom workers, queues, cron jobs and webhooks.
Telegram, WhatsApp, VK, web widgets. From a simple menu to an AI sales bot with payments, a user account, an admin panel and analytics. TON Mini Apps and Web App interfaces.
Design and code with no middlemen. Next.js / React, animations, 100/100 performance, SEO, CMS, payments, cart, inventory and 1C integration. A landing page that sells rather than just looking pretty.
Smart contracts in Solidity, Tact and Tolk. Token launches, NFT collections, staking, vesting, DEX listings, wallets and Telegram Mini Apps on TON.
iOS and Android, native or cross-platform. Publishing to the App Store and Google Play, push notifications, offline mode, biometrics, subscriptions and on-device AI.
Detection, segmentation, OCR, face and licence plate recognition, quality control on production lines, real-time video analytics — in the cloud or right on site.
Fine-tuning and LoRA, distillation, custom datasets, embeddings, recommender systems, forecasting and classification. Deployed on your own servers, with no data leaving your perimeter.
High-load APIs, microservices, Docker, CI/CD, monitoring. Plus 1C integrations and customisation — a rare combination that covers the whole corporate stack.
This list is not exhaustive — it is simply what people order most often. Tell me what you need: I will dig into it and explain how to solve it.
Message me on TelegramWrite a couple of lines about your project and I will come back with an approach, a timeline and a budget range. Free, and without calls for the sake of calls.
I pick the tool for the task, not the task for the tool.
My main focus. Not "bolting ChatGPT on", but designing a system that reliably solves your problem.
An agent with access to your tools: database, CRM, calendar, email, payments, browser. It plans, calls functions, checks itself and escalates to a human. Every step is logged, with protection against hallucinations built into the architecture.
We load your policies, contracts, knowledge base and correspondence, and get an assistant that answers from facts and cites the source. Hybrid search, reranking and chunking tuned to your document types.
Object and defect detection, counting people and goods, OCR of documents and receipts, plate recognition and tracking. Runs in the cloud or on modest hardware on site.
Call and meeting transcription, diarisation, voice bots, speech synthesis, automatic summaries and tagging. Sales call analysis with script scoring.
I collect and label the dataset, fine-tune (LoRA or full), quantise and deploy self-hosted. Your data never reaches vendors and inference costs drop several times over.
Recommender systems, scoring, anti-fraud, demand and churn forecasting. Plus dashboards, so the result is visible and not just felt.
Pick a task and press Run — I will show step by step what happens inside. This is an illustration of the logic, not a live model run.
Press Run to see the chain of reasoning and tool calls.
From token idea to listing and a living community. EVM networks and TON.
ERC-20, BEP-20, Jettons on TON. Tokenomics, fees, anti-bot, blacklists, mint and burn, team vesting and presale contracts.
Solidity, Tact, Tolk. Staking, farming, vesting, escrow, lotteries, multisig. Tests, simulations, gas optimisation and an audit checklist before deployment.
Artwork generation, metadata, IPFS, a mint site with whitelist, royalties, marketplace integrations and dynamic NFTs.
Web apps with wallet connection, signing, swaps and charts. TON Connect, WalletConnect, EVM providers.
Games, tappers, wallets and stores right inside Telegram. Payments in TON and Stars, referral systems and leaderboards.
Sniper bots, transaction trackers, arbitrage scanners, indexers, nodes and Telegram alerts.
I build apps worth shipping: a tidy interface, smooth animation, offline mode and clean architecture inside. I handle the whole path — from mockup to store release and updates.
No bureaucracy. Five steps from a Telegram message to a working product.
You message me on Telegram and describe the task in your own words. I ask questions and suggest options. Often it becomes clear right here that it can be done simpler and cheaper.
I lock down scope, timeline and cost, then split it into stages with clear deliverables — you see progress instead of "almost there".
A first working version quickly — something to try rather than documentation to read. We correct course early, while it costs nothing.
I write, test and demo at every stage. You know what is happening, and the code is yours from day one.
Deployment, monitoring and a guide for your team. After that — support and new features as the product grows.
Three ways to work together. Which one fits we decide in the first conversation, free of charge.
When the task is clear and you need a finished result by a deadline.
When the work is exploratory and the scope cannot be measured upfront.
When the product is already live and needs to keep moving.
I deliberately keep prices off the site: an honest estimate is calculated for the task, not taken from a price list. Message me and I will give you a range in my first reply.
Do not see your question? Just write. I answer quickly and to the point.
Ask a question→It depends on scope. A simple bot or landing page starts at a few hundred dollars; an AI system or an app is estimated stage by stage. After the first conversation I give a range, after the brief an exact figure. I work with staged prepayments tied to milestones.
A landing page takes 3–7 days. A bot, a week or more. An AI agent or an app, 2–6 weeks depending on complexity. I take urgent work out of turn when it is genuinely feasible — I will not promise the impossible.
I do most of it myself — it is faster and nothing gets lost in translation. On larger projects I bring in people I trust: a designer, a QA engineer, a second backend developer. Responsibility and communication stay with me.
You do, entirely, from the first commit. Repository, access, servers, keys — everything is handed over. No lock-in to my infrastructure and no "support subscription" without which things fall apart.
That is fine, and even convenient: you describe the task in plain language, I translate it into a technical solution and explain the options without jargon. The choice is always yours, with clear pros and cons.
For the first period after delivery I fix everything under warranty. After that it is up to you: one-off changes or monthly support with monitoring and updates. I do not disappear.
If that matters, I deploy models on your own servers and nothing leaves your perimeter. Where cloud is acceptable, I use modes that exclude training on your data and strip personal information before sending.
Those are my favourite. If it is technically possible, I will do it. If not, I will say so straight away and suggest a workaround rather than burning your budget.
Describe your task in two sentences — I will pull out the rest with questions. The consultation is free and I reply within 15 minutes.
Message me on Telegram