Browse categories
Explore
Fiverr Pro
English
$
USD
I will build a rag ai chatbot that answers from your pdfs with source citations
Sergii M
About this gig
Please message me before ordering - tell me the use case in two lines.
Tried an AI chatbot and it made things up? Most AI features break on real data: invented answers, blank outputs when the API fails, no way to check a result.
I add AI to your product the careful way: answers come from your own documents, each with a citation you can open, and the code handles API failures instead of crashing.
What you get: LLM integration (OpenAI, Claude or others), retrieval over your documents (RAG), tests and error handling, clean Python code.
Why me: 12+ years in IT; a tender-analysis SaaS on Claude API in production since 2025; an LLM gateway with per-team budgets on AWS Bedrock.
Not a fit if you need a no-code chatbot template.
Get to know Sergii M
Sergii M
Python AI developer, over 12 years in IT, document AI and RAG in production
- FromUkraine
- Member sinceDec 2025
- Avg. response time1 hour
Languages
English, Ukrainian
12+ years in IT, founder of Cloverity. Up to 98% accuracy extracting SEC 10-K/10-Q statements from PDFs. A 44K-line PHP backend moved to Python FastAPI with 918+ tests.
I build Python and AI systems that read hard documents (contracts, tenders, financial filings) and return answers you can check, with citations.
In production, not demos: a tender-analysis SaaS, live since 2025.
You own the code and the data: your repository, your server, NDA on request.
Send me 2-3 sample documents and I will tell you what is possible.
My Portfolio
FAQ
Why should you choose me?
I build AI assistants that can be checked: every answer cites its source, API failures are handled, and the code is tested. A tender-analysis SaaS on Claude API has run in production since 2025. 12+ years in IT.
What's included?
The package deliverables, full source code in your repository, Docker setup, and a short written handover. Premium also includes 30 days of bug fixes and small changes after delivery.
What's not included?
LLM API fees, hosting costs, content writing, and training custom models. The document set and the questions it should answer are agreed before we start.
Which AI providers do you use?
OpenAI, Anthropic Claude, AWS Bedrock or others, chosen for your case and budget.
How do I know the answers are real?
Every answer shows the passage it came from, so you can open the source and check. If nothing relevant is found, the assistant says so instead of guessing. The pilot includes a check on questions you provide.
What happens after the pilot?
You get a written report on how the assistant answered your questions and a fixed quote for the full build. If the pilot shows RAG is not the right tool, I tell you that too.
Is my data safe?
Yes. I'm happy to sign your NDA before you get me access to your code or data. I use that access only for this order and remove it after delivery.
Where do my documents actually go?
They stay in storage you control: your server or your cloud account. The only outside call is to the LLM provider you choose, and I show you exactly what text is sent to it.
Do I need my own API keys, and what will it cost to run?
Yes, keys and accounts are in your name, so you own them and pay the provider directly. The running cost depends on document volume and questions per month; you get an estimate in the pilot report.
Can the knowledge base stay up to date when documents change?
Yes. New or changed documents are re-indexed, so answers follow the latest version. How it is triggered (upload, schedule or a watched folder) is agreed in the order.
