Dieses Stellenangebot ist nicht mehr verfügbar
Über
Scope of Work:
We have a collection of 15-20 books (with more to be added later) that are broken down into paragraphs and stored in a PostgreSQL database. The schema for the Books table is as follows:
Table: Books
Column Name Data Type Description
chapter_title text Title of the chapter
para text Paragraph content
book_name text Name of the book
Your task is to create a RAG-based system that can efficiently answer a wide range of user queries. The queries will vary in scope, including but not limited to:
Specific facts: Retrieving a specific fact from a certain line or paragraph.
Chapter-based: Answering queries related to an entire book or specific chapter.
Multiple chapters/books: Handling questions that span across multiple chapters of the same or different books.
Key Features and AI Considerations:
We have conducted substantial research and have received suggestions, including the following AI-driven recommendations:
Feature Engineering: Creating a "concepts" column for better contextual understanding.
GraphRAG Approach: Exploring the use of Graph-based RAG rather than a simple RAG pipeline for improved performance.
Language Consideration:
Most of the books are in Urdu, with a subset in English. The system should be able to handle multilingual queries effectively.
Deliverables:
Implementation: Develop the RAG system.
Evaluation and Benchmarking: Performance evaluation and fine-tuning of the system to ensure it meets the required query accuracy and speed.
Proposal Requirements:
A detailed proposal in PDF format, emphasizing the system architecture.
Share relevant experience with similar RAG projects, particularly those that involve multilingual data.
Evaluation Criteria:
Does your proposal surpass the AI-generated approach?
Does it reflect your deep expertise in AI, NLP, and RAG systems?
How innovative and scalable is your proposed solution?
Additional Notes:
If you have experience working on similar retrieval-augmented systems or NLP-based knowledge retrieval projects, especially with multilingual support, please include links to live or past projects as a plus.
We look forward to your detailed proposal
Contract duration of 3 to 6 months. with 40 hours per week.
Mandatory skills: Python, API, PostgreSQL, Graph Database, Vector Database, Retrieval Augmented Generation, Generative AI, Natural Language Processing
Sprachkenntnisse
- English
Hinweis für Nutzer
Dieses Stellenangebot wurde von einem unserer Partner veröffentlicht. Sie können das Originalangebot einsehen hier.