Browse categories
Explore
Fiverr Pro
English
$
USD
I will build a rag chatbot trained on your business documents
Arijit K
About this gig
I will build a focused RAG knowledge assistant that retrieves from your approved business documents and returns answers with source references. Useful for policies, support knowledge, technical documentation and operational playbooks. RAG retrieves context; model fine-tuning is not included.
Basic: up to 25 text documents or 500 pages, whichever comes first; one model/vector-store configuration, RAG API/local demo with model integration, document citations, source, README and 10 acceptance questions. Standard: up to 100 documents/2000 pages, FastAPI, vector database, basic auth, Docker, ingestion/evaluation workflow and handover. Premium: up to 500 documents/10000 pages, role-aware access, background ingestion, logging, buyer-owned deployment and one defined app integration.
No custom frontend, SSO, scans/OCR, complex-table extraction, security certification or ongoing support. Buyer pays model, cloud and database fees.
LLMs can be wrong. I include grounding, evaluation and documented limits. Message before ordering with formats, volume, users, hosting and success criteria.
Get to know Arijit K
Arijit K
AI Engineer RAG Workflow Automation FastAPI Backends
- FromIndia
- Member sinceJul 2025
- Avg. response time2 hours
Languages
English, Hindi
AI and backend engineer with a data science and ML background from NSUT (ECE, AI/ML). I build practical AI systems that automate business workflows: RAG knowledge assistants, LLM integrations, Python/FastAPI APIs, and ML pipelines. My experience includes commerce infrastructure at Nirmit Bharat (DigiHaat/ONDC) and ML engineering work with the Indian Air Force. I focus on grounded AI behavior, clear scope, maintainable code, and documented handovers. Message me with your objective, data/API availability, and definition of done
FAQ
Which documents and citations are supported?
Text PDFs, DOCX, TXT, Markdown, CSV and HTML, subject to sample review and your permission to share them. Citations depend on extraction quality and can identify document, page, section or chunk. Scans and complex tables need a separate preprocessing scope.
Can you guarantee correct answers, and are running costs included?
No. I reduce errors with retrieval grounding, citations, evaluation questions and refusal behavior. Buyer-owned model APIs, hosting and vector-database charges are separate. Application integration is included in Premium; custom UI and SSO require separate scope.
