I will build a rag chatbot app on your data with pgvector and langchain


Level 1
Vetted by Fiverr Pro
Junaid Aslam was selected by the Fiverr Pro team for their expertise.
About this gig
Please message me to discuss requirements before placing the order.
Need a RAG chatbot that answers from your PDFs, Word files, wikis, and support emails? I build a retrieval augmented generation (RAG) chatbot app that reads your documents and answers with source citations.
What I build
- Document ingestion: PDF, Word, Excel, Markdown, web pages, and plain text
- Chunking, embeddings, and semantic search, tested on your sample questions
- Embeddings in PostgreSQL with pgvector, so you add no new vector database bill. Qdrant or Pinecone if you prefer
- LangChain.js or Semantic Kernel with OpenAI, Claude, Gemini, or open models (Llama, DeepSeek, Qwen) via Ollama or OpenRouter
- Answers with sources. If the answer is not in your files, the bot says so
- Chat widget for your website (React, Next.js, or JavaScript) or an API (ASP.NET Core or Node.js)
- Premium: admin upload page, user logins, and WhatsApp or Telegram
- Docker deploy to your server, AWS, or Azure, in an EU, UK, or US region
- Source code and a handover guide
Send me a sample of your documents and tell me who will use the bot. I will tell you if RAG fits your case and what it costs to run.
Let's discuss first.
Get to know Junaid Aslam
ASP Dot NET Core, Web API, React, JS, TS, SQL, NoSQL, AWS, Azure, AI Chatbot RAG
Level 1
Junaid Aslam is part of the Fiverr Pro catalog and has been hand-picked by a dedicated Fiverr Pro team for their skills and expertise.
Vetted for
Software Development
- FromPakistan
- Member sinceSep 2015
- Avg. response time1 hour
- Last delivery1 month
Languages
Urdu, English
My Portfolio
Other AI Development Services I Offer
FAQ
What files can the bot read?
PDF, Word, Excel, web pages, and plain text. Scanned PDFs need an extra OCR step, so tell me before you order.
Will it make things up?
It answers from the passages it finds and shows the source. If it finds nothing relevant, it says so. I cannot promise zero errors, so I test it with your own questions.
Why pgvector?
It keeps your vectors inside Postgres. If you have no Postgres, I set one up. Qdrant and Pinecone are options.
Who pays for AI usage?
You do, through your own API key. A local model has no per-question cost.
Can it run on my own server?
Yes. With a local model through Ollama, nothing leaves your server.
