Skip to main content

USP – Educational RAG Application

AI/ML Developer

Retrieval-Augmented Generation application for Universidade de São Paulo enabling intelligent question-answering for educational content.

AI / ML

Context

USP needed intelligent system allowing students and faculty to interact conversationally with vast academic materials, lecture notes, and research papers.

Challenge

Building accurate question-answering system with document processing pipelines, semantic search, and optimized retrieval for educational queries.

Solution

Implemented LangChain with large language models and vector databases (Pinecone/Weaviate). Built document processing with AWS Textract and React-based natural language search interface.

Technology Stack

ai

LangChainAWS BedrockLarge Language ModelsNLP

processing

AWS TextractDocument ProcessingCustom Workflows

search

PineconeWeaviateVector EmbeddingsSemantic Search

frontend

ReactNatural Language Interface

Impact & Results

Intelligent question-answering with conversational interaction for educational queries

Document processing pipelines for academic materials and research papers

Advanced embedding strategies improving response relevance by 40%

Natural language search interface for students and faculty

Technologies

LangChain
AWS Textract
AWS Bedrock
Pinecone
React
Python
Vector Databases

Links

More Projects

Enterprise

Production URL Shortener – MLH PE Hackfest 2026

Production Engineer

Production-grade URL shortener built for the MLH Production Engineering Hackathon, demonstrating SRE practices: horizontal scaling, Redis caching, observability, load testing, and incident runbooks.

View Project
Blockchain / Web3

Learnway – Blockchain Education Platform

Fullstack Developer

Full-stack platform for a blockchain-based educational application on Lisk blockchain that gamifies learning through point-based rewards and token incentives.

View Project

Interested in similar work?

Let's discuss your project requirements