You will get a RAG document Q&A bot that answers from your files with citations
Rising Talent
Project details
Your documents become a bot that answers questions and points to the exact source — the file and the page — so you can trust every answer. No made-up facts: if the documents don't cover it, it says so.
It handles long PDFs, scanned reports, spreadsheets, and many files at once. I take your sample documents, build the retrieval and answer pipeline, and show it answering your real questions with citations before you pay.
Built with Python, LangChain or LlamaIndex, and a vector database. Everything in writing — no calls. You get the code, a short setup guide, and revisions until the answers are right.
Good fit for: contracts and policies, research and reports, support and product docs, internal knowledge bases, and any pile of files you keep searching by hand.
It handles long PDFs, scanned reports, spreadsheets, and many files at once. I take your sample documents, build the retrieval and answer pipeline, and show it answering your real questions with citations before you pay.
Built with Python, LangChain or LlamaIndex, and a vector database. Everything in writing — no calls. You get the code, a short setup guide, and revisions until the answers are right.
Good fit for: contracts and policies, research and reports, support and product docs, internal knowledge bases, and any pile of files you keep searching by hand.
AI Algorithms
Large Language Model, Transformer ModelAI Applications
Conversational AI, Natural Language Generation, Natural Language UnderstandingAI Development Language
PythonAI Models
BERT, ChatGPT, GPT-4What's included
| Service Tiers |
Starter
$150
|
Standard
$350
|
Advanced
$700
|
|---|---|---|---|
| Delivery Time | 3 days | 5 days | 8 days |
Number of Revisions | 2 | 3 | 5 |
AI Model Integration | |||
Batch Normalization | - | - | - |
Database Integration | - | ||
Detailed Code Comments | |||
Image Upscaling | - | - | - |
MLOps | - | - | - |
Model Deployment | - | - | |
Model Documentation | - | ||
Model Monitoring | - | - | |
Model Testing & Optimization | - | - | |
Model Tuning | - | - | - |
Natural Language Processing | |||
NLP Tokenization | - | ||
Pre-Training | - | - | - |
Prompt Engineering | |||
Setup File | |||
Source Code |
Frequently asked questions
About Bogdan
Senior Claude Code & MCP Engineer | Clinical AI, RAG, Multi-Agent
Kyiv, Ukraine - 11:31 am local time
13 years as a clinical psychologist before AI engineering — rare combination. I build agent systems that pass both engineering review and clinical scrutiny, because I've sat on both sides.
— What I build —
• MCP servers exposing internal tools to Claude Desktop / Claude Code — auth, rate-limiting, error paths, prompt-injection guards. Not generic forks; tuned to each client's workflow.
• Multi-agent systems: dispatcher → 3-7 specialized subagents → quality-gate layer → handoff protocol before context-compaction. Persistent memory, not prompt-stuffing.
• Hook + sandboxing layers: pre-action guards, post-action telemetry, watchdog daemons. The boring guarantees that make AI systems reliable instead of theatrical.
• Healthcare / clinical AI: PHI deidentification with crypto envelope chains, supervisor-grade audit logs, clinical formulation modules grounded in CBT/Schema/MAPS-track frameworks. Where most AI engineers don't have clinical context to know what to refuse.
— Recent work (anonymized where NDA) —
• EU clinical SaaS extension: 3-role architecture (patient / therapist / supervisor), 11,309 production lines across 8 files, full code-review 10+ rounds.
• B2B research pipeline: 430 manually verified contacts across 12 markets in 2 weeks. Custom triangulation + Airtable delivery. over 2x cold-database response rate.
• Open Claude Code reference implementation: 20+ specialized agents, MCP integrations, memory layer, hook system, bridge architecture. Production-grade, used as my own daily ops for 12+ months.
— Stack —
Anthropic Claude (Fable 5 / Opus 4.8 / Sonnet 4.6 / Haiku 4.5), MCP (stdio + HTTP), LangGraph, RAG (ChromaDB / vector DBs), Python, TypeScript, Node.js, n8n.
— How I work —
First engagement = fixed-price milestone: scope doc (day 1-2), working system (day 5-10), production handoff (day 12-14). One revision. Daily git pushes — you audit progress async.
Bilingual EN/UA/RU.
— CTA —
Send your use case (workflow + pain point + stack). I reply within 24h on weekdays with yes/no + rough scope, or a referral if not my fit.
Steps for completing your project
After purchasing the project, send requirements so Bogdan can start the project.
Delivery time starts when Bogdan receives requirements from you.
Bogdan works on your project following the steps below.
Revisions may occur after the delivery date.
You share sample docs and questions
You send a few sample documents and the kinds of questions you need answered.
I build it and demo on your docs
I build the retrieval and answer pipeline and show it answering your real questions with citations.