You will get Custom AI Agents, RAG Systems & Automation | Python, LLMs, MCP Servers


Project details
You get an AI system that actually works in production — not a chatbot demo, not a Jupyter notebook, but deployed, tested code that connects to your real data and does useful work from day one.
I build AI agents that reason, use tools, and take action autonomously. RAG systems that answer questions over your documents with cited sources. MCP servers that give Claude real-time access to your APIs and databases. Workflow automations that replace hours of manual work with a single trigger.
My background is 13 years in financial markets plus hands-on product leadership in crypto. I've built ML-powered trading pipelines, LLM sentiment analyzers, and autonomous agents that execute real transactions. This means I understand not just how AI works, but how it fails — and I engineer around those failure modes.
Every deliverable ships with Docker deployment, source code, documentation, and a test suite. I work in milestones with daily updates — you always know where things stand.
Whether you need a Telegram bot powered by Claude, a RAG system over thousands of documents, or a multi-agent pipeline that monitors and acts — I deliver production-ready AI you can trust.
I build AI agents that reason, use tools, and take action autonomously. RAG systems that answer questions over your documents with cited sources. MCP servers that give Claude real-time access to your APIs and databases. Workflow automations that replace hours of manual work with a single trigger.
My background is 13 years in financial markets plus hands-on product leadership in crypto. I've built ML-powered trading pipelines, LLM sentiment analyzers, and autonomous agents that execute real transactions. This means I understand not just how AI works, but how it fails — and I engineer around those failure modes.
Every deliverable ships with Docker deployment, source code, documentation, and a test suite. I work in milestones with daily updates — you always know where things stand.
Whether you need a Telegram bot powered by Claude, a RAG system over thousands of documents, or a multi-agent pipeline that monitors and acts — I deliver production-ready AI you can trust.
What's included
| Service Tiers |
Starter
$800
|
Standard
$2,000
|
Advanced
$5,000
|
|---|---|---|---|
| Delivery Time | 5 days | 10 days | 21 days |
Number of Revisions | 2 | 3 | 5 |
Number of Model Variations | 1 | 2 | 3 |
Number of Scenarios | 1 | 3 | 5 |
Number of Graphs/Charts | 0 | 1 | 3 |
Model Validation/Testing | - | ||
Model Documentation | - | ||
Data Source Connectivity | |||
Source Code |
Frequently asked questions
About Dumitru
Web3 & AI Engineer | DeFi, Solidity, Python | Ex-Crypto Product Owner
Chisinau, Moldova - 7:24 am local time
I was Product Owner of a large crypto project — from zero to launch, tokenomics to smart contract architecture, managing dev teams while personally shipping critical infrastructure. I understand both sides: what the business needs and how the code must work to not lose funds.
Smart contract security
I audit Solidity contracts for reentrancy, access control, oracle manipulation, front-running, precision loss, and economic exploits. I don't just find bugs — I understand the financial logic behind the protocol, which lets me catch vulnerabilities that pure-code reviewers miss. Recent audit: found critical issues in a contract that passed two prior reviews.
Trading systems & execution infrastructure
I design and build production trading bots for CeFi and DeFi venues:
Polymarket penny sniping bot with Chainlink oracle consensus, CLOB order signing, and Telegram control panel.
Exchange integrations: Binance, Bybit, OKX, Interactive Brokers, Polymarket (Web3)
Signal strategies: mean reversion, momentum, arbitrage, market making, ML-enhanced signal fusion
Full lifecycle: backtesting → shadow trading → live execution → monitoring
AI & agent development
I build AI-powered tools for crypto and trading — from LLM-driven sentiment analysis to MCP servers that give Claude and other AI agents real-time access to market data. I work with scikit-learn, PyTorch, Hugging Face transformers, and LangChain for custom model pipelines.
Tech
Solidity · Python (asyncio, web3.py, ccxt, FastAPI) · Node.js · MQL4/5 · Docker · PostgreSQL · Redis · AWS
How I work
Fixed-price milestones through Upwork escrow. Each deliverable is tested, deployed, and documented. I communicate in advance when something will take longer than expected — no silent delays. Post-delivery support included.
Steps for completing your project
After purchasing the project, send requirements so Dumitru can start the project.
Delivery time starts when Dumitru receives requirements from you.
Dumitru works on your project following the steps below.
Revisions may occur after the delivery date.
Scoping and architecture
I analyze your requirements, select the right model and tools (Claude, OpenAI, open-source), design the system architecture, and share a technical plan for your sign-off
Core build and integration
Build the AI agent or RAG pipeline, connect your data sources, implement tool use and memory. You get access to a private repo with progress updates