You will get RAG Knowledge Base — AI That Answers from Your Documents
Rising Talent
Project details
Most AI chatbots make things up when they don't know the answer. This system doesn't.
I build RAG systems that answer questions exclusively from your own documents — contracts, manuals, reports, knowledge bases — and cite the exact source for every response. If the answer isn't in your files, the system says so instead of guessing.
The result is a production-ready AI assistant your team can actually trust: accurate, traceable, and built on your content.
What makes this different from a generic chatbot build: I've shipped this architecture in production. Licitei's AI layer — a LangGraph agent with 8 specialized retrieval tools — indexes 16,917 records and returns source-cited responses in real time. Same core system, adapted to your documents and domain.
Every tier includes Python source code, setup instructions, and a working demonstration against your actual files. You own everything delivered.
I build RAG systems that answer questions exclusively from your own documents — contracts, manuals, reports, knowledge bases — and cite the exact source for every response. If the answer isn't in your files, the system says so instead of guessing.
The result is a production-ready AI assistant your team can actually trust: accurate, traceable, and built on your content.
What makes this different from a generic chatbot build: I've shipped this architecture in production. Licitei's AI layer — a LangGraph agent with 8 specialized retrieval tools — indexes 16,917 records and returns source-cited responses in real time. Same core system, adapted to your documents and domain.
Every tier includes Python source code, setup instructions, and a working demonstration against your actual files. You own everything delivered.
AI Algorithms
Large Language Model, Transformer ModelAI Applications
AI Chatbot, AI Mobile App Development, AI Text-to-Speech, Conversational AI, Natural Language Generation, Natural Language UnderstandingAI Development Language
PythonAI Tools
GitHub Copilot, Gradio, Hugging Face, StreamlitAI Models
ChatGPT, GPT-4, LLaMA, OpenAI Codex, WhisperWhat's included
| Service Tiers |
Starter
$200
|
Standard
$350
|
Advanced
$600
|
|---|---|---|---|
| Delivery Time | 7 days | 10 days | 18 days |
Number of Revisions | 1 | 2 | 2 |
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 |
Optional add-ons
You can add these on the next page.
Additional Document Collection
(+ 2 Days)
+$80
Cloud Deployment
(+ 3 Days)
+$80
External Platform Integration
(+ 2 Days)
+$100Frequently asked questions
About Vyktor
AI Engineer | LLM Integration & Intelligent Agents | Python
Jaboatao dos Guararapes, Brazil - 4:54 pm local time
Recent work:
LICITEI — ETL + LLM Agent for Public Procurement (Python, FastMCP, LangGraph)
Built two integrated systems for a platform that helps Brazilian micro-entrepreneurs find government contracts:
ETL pipeline: 12,050 records processed in a single run from a live government API (0 invalid, 0 DLQ), with full pagination, exponential backoff, idempotent upsert in MongoDB Atlas, and Medallion architecture (Kafka KRaft, Bronze/Silver/Gold with PyIceberg ACID + time travel). 36,385 total in Bronze, 16,917 unique contracts in Silver/Gold — 95.2% MEI-eligible. Orchestrated with Prefect, monitored via Streamlit dashboard.
LLM agent server: FastMCP with LangGraph ReAct agent, 8 tools (procurement search, document summary, compliance checklist, CNAE lookup, and more), Groq API (llama-3.3-70b) as primary LLM with automatic fallback to local Ollama, per-session memory via MemorySaver, and SSE streaming endpoint. Users query government procurement data in plain language.
Managed a team of 8 engineers across 5 functional tracks.
KEIZA CAPITAL — Blockchain Fintech (Co-founder)
Co-founded a fintech connecting Brazil's PIX to stablecoins for international payments. Processed R$30,000+ during MVP with a flat 2% fee (vs. up to 7.5% traditional). Partnership with Avenia for liquidity and compliance.
What I build:
✔ RAG systems connected to your data (documents, databases, APIs)
✔ LLM agents with tool use, memory, and structured output
✔ FastAPI + FastMCP + Python backends for AI applications
✔ ETL pipelines feeding AI systems
✔ MCP servers for LLM-tool integration
Stack: Python · PySpark / Apache Spark · FastAPI · LangChain · LangGraph · FastMCP · RAG · ChromaDB · Google Gemini · Groq API · MongoDB Atlas · Kafka · PyIceberg · Prefect · Pydantic v2 · pytest · Docker · GitHub Actions
Systems Analysis & Development graduate, CESAR School (Recife, Brazil).
If you have a concrete problem that needs AI to actually work in production — message me describing it. I'll tell you in 24 hours if I can solve it and how.
Steps for completing your project
After purchasing the project, send requirements so Vyktor can start the project.
Delivery time starts when Vyktor receives requirements from you.
Vyktor works on your project following the steps below.
Revisions may occur after the delivery date.
Document Ingestion
Your files are processed, chunked, and indexed into a vector database. You receive confirmation of how many documents were successfully indexed.
RAG Pipeline Build
Retrieval logic and LLM integration are configured. The system is tested internally against your documents to validate accuracy and citation behavior.