Kategorien durchsuchen
Entdecken
Fiverr Pro
Deutsch
$
USD
I will build a rag chatbot trained on your documents
Adam M
Über diesen Service
Your docs, not generic chat.
I build RAG chatbots that answer from YOUR PDFs, websites, or databases - with citations. Chunking, embeddings, a vector store, and a grounded chat UI. You get source code, a runnable app, and a short handoff.
What you get:
- Ingest PDFs, sites, docs, or a DB dump
- - Chunking matched to your content
- - Embeddings + vector DB (Pinecone, Chroma, Qdrant, Weaviate, or pgvector)
- - Dense / hybrid search and optional reranking
- - Chat API + web UI with source citations
- - Prompts that refuse to invent facts when retrieval is empty
Stack: Python, FastAPI or Next.js, LangChain or LlamaIndex, OpenAI / Anthropic / local LLMs, Docker.
Process: you send sample docs + real questions, I ship an index + chat loop, we iterate on wrong/missing answers, then handoff.
To start: 5-20 sample files and 10 real questions. Message me if you need Slack, SSO, or a specific vector DB.
Lerne Adam M kennen
Adam M
AI Engineer, RAG, ML and Agents
- AusMarokko
- Mitglied seitSept. 2025
- ⌀ Antwortzeit1 Stunde
Sprachen
Englisch, Französisch, Arabisch
I build production AI systems: RAG chatbots grounded in your docs, custom ML and deep learning models, generative AI apps, and tool-using agents. Python, LangChain, LlamaIndex, PyTorch, FastAPI. You get source code, evaluation, and a handoff - not a wrapped chatbot template. I will tell you if RAG, a model, or an agent is the wrong tool. Based in Morocco. English and French.
FAQ
Will the bot invent answers not in my documents?
RAG still uses an LLM, so it can drift. I ground answers in retrieved chunks, show citations, and prompt it to say it does not know when retrieval is weak. Premium adds an eval set so we measure that.
Can you keep my data off OpenAI?
Yes - Azure, a VPC endpoint, or a local model (Ollama / vLLM). Say this up front. I will not send your files to a public API unless you approve it.
What if my PDFs are scanned or messy?
Scanned PDFs need OCR. Tables and multi-column layouts need extra parsing. Tell me in the order; OCR or heavy cleanup may be an extra if volume is large.
Do I own the code?
Yes. You get the project files. Third-party APIs stay on your accounts and billing. I do not claim your data or lock you into a hosted product of mine.
