Team
Morad.IA
This team is at maximum capacity.
Project Concept
Morad.IA: Democratizing Homeownership
Morad.IA is an AI-powered virtual concierge that democratizes access to homeownership by transforming complex real estate bureaucracy into a simple, automated WhatsApp conversation for informal and low-income workers.
The Problem: A Bureaucratic Nightmare
Real estate credit (such as the Minha Casa Minha Vida program) is a bureaucratic maze of “negative certificates,” income proofing, and complex legal rules. This disproportionately affects low-income families and informal workers who lack the specialized knowledge or documentation required to navigate the system, often leading to immediate rejection after weeks of waiting in line.
The Solution: A Multi-Agent AI Concierge
Accessible through familiar tools like WhatsApp, Morad.IA utilizes a multi-agent AI architecture to guide users through the entire pre-approval process:
- Intelligent Onboarding: Users communicate via simple audio or text (e.g., “I’m a mechanic earning 4k a month and want a 200k house”).
- Multimodal Validation: The AI performs real-time OCR on photos of documents (IDs, bank statements), identifying expired or illegible files instantly to prevent submission errors.
- Smart Simulation: By cross-referencing user data with real-world banking rules (via RAG) and simulating public registry queries, it generates a “Structured Dossier” ready for official bank submission.
Social Impact
- Financial & Housing Inclusion: It empowers informal workers and the self-employed by translating dense legal jargon into everyday language, providing them with the same expert guidance typically reserved for high-income earners.
- Reducing Inequality: By automating the “pre-qualification” phase, Morad.IA lowers the barrier to entry for government housing programs, directly addressing the national housing deficit.
- Efficiency for the Underserved: It eliminates the costly and time-consuming “back-and-forth” at banks and registries, preventing citizens from wasting limited resources on applications destined for rejection due to simple documentation errors.
Entry
Status: Submitted
Last saved: April 12 at 4:51 PM -03
Team Roster (team is at max capacity)
Message board not available for this team yet.
Renzo Real Machado Filho Team Lead RSVP Approved
Student at USP
Responsibility: AI Orchestration, Data Gathering, and RAG Implementation.
Agent Logic: Orchestrated the multi-agent workflow using LangGraph, including the "Traffic Guard" router that decides between general conversation or technical research.
Data Collection: Acted as the "Data Miner," writing scripts to scrape and filter housing information from official sources.
Integration: Connected the legal knowledge base to the agents, ensuring the chatbot uses LangChain to retrieve context from ChromaDB.
I am a Computer Science undergraduate at IME-USP and a current FAPESP research fellow dedicated to bridging theoretical mathematics with applied AI. I also lead data-driven initiatives to analyze urban infrastructure, contributing to the open-source sideseeing-tools library to support public policy development. With a background that includes managing university-scale GNU/Linux network infrastructure and winning 3rd place at the 6th FGV Digital Currency Datathon for deep learning time-series forecasting, I offer a unique blend of academic rigor and systems-level expertise. I am a collaborative builder specifically interested in exploring technical architecture and forming strategic partnerships to develop innovative, data-driven solutions.
AI, computer vision, optical flow, sidewalk analysis, image processing, machine learning, Python development, Linux system administration, automated deployment scripts.
My work includes FAPESP-funded research at IME-USP developing computer vision tools for education and urban analytics through the sideseeing-tools library. I am also refining my full-stack capabilities by building scalable AI systems using FastAPI and Docker. Currently, I lead the technical architecture for "Ajudaí", a domestic services marketplace developed within a startup simulation course focused on market validation.
Felipe Cordeiro Caram RSVP Approved
Undergrad Student at Instituto de Matemática, Estatística e Ciência da Computação da USP
Dropped.
Analytical problem-solver and Computer Science student at IME-USP. Passionate about the intersection of technology, strategy, and finance. Skilled in quantitative modeling and deconstructing business models to identify strategic growth opportunities. Seeking to leverage data-driven insights to bridge technical and business objectives, scale operations, and drive measurable value.
Highly interested in the intersection of technology, finance, and corporate strategy. Seeking to connect with professionals across financial services, consulting, and the tech ecosystem to deepen my understanding of value creation and market dynamics. I am eager to explore diverse roles where I can leverage my quantitative background and operational insights to solve complex business challenges and scale impactful solutions.
Currently developing the MVP for "Ajudaí," a marketplace startup for domestic services in Brazil, focusing on business strategy, visual identity, and operational structure using AI development tools. Concurrently, I am expanding my technical and analytical foundation through rigorous academic projects focused on operating systems, full-stack web development, and machine learning applied to finance.
JONATHAN DA SILVA PEREIRA RSVP Approved
Student at USP
Responsibility: Infrastructure, User Experience, and Frontend.
Infrastructure: Configured the Twilio API and ngrok to create a public tunnel for real-time WhatsApp connectivity.
API Management: Developed the FastAPI backend to handle incoming webhooks and manage asynchronous media downloads (like photos of documents).
Dashboard: Built the broker-facing dashboard using Streamlit, which reads banco.json to display a user's dossier in real-time.
Jonathan Pereira is an Economics Student and AI member at FEA.dev, based in São Paulo, Brazil. He is pursuing a Bachelor's degree in Economics from the School of Economics, Business and Accounting at the University of São Paulo (FEA-USP) and also studied 2 years of Computer Science at the Instituto de Matemática, Estatística e Ciência da Computação (IME-USP). His work involves analyzing systemic risk and market contagion using statistical models and dimensionality reduction techniques. He is open to introductions and prefers to be contacted via email.
Systemic risk analysis, market contagion, statistical modeling, time-series analysis, dimensionality reduction, artificial intelligence, quantitative workflows, data science, financial dependencies, economics, computer science.
Analyzing systemic risk and market contagion at FEA.dev using Monte Carlo simulations, Granger causality tests, and Principal Component Analysis (PCA). As a member of the Artificial Intelligence nucleus, he implements statistical models and dimensionality reduction techniques to evaluate financial dependencies. Additionally, he contributes to the data nucleus at IME Jr, leveraging a background in Computer Science and Economics to execute quantitative workflows.
Gabriel Freire Ushijima RSVP Approved
University Student at Universidade de São Paulo
Responsibility: Computer Vision, Structured Data, and Financial Logic.
Vision AI: Developed the extrair_dados_do_documento tool using the Gemini 2.5 Flash API to perform OCR on IDs and pay stubs, enforcing Structured Output via JSON schemas.
Mathematics: Programmed the financial engine responsible for calculating the 30% installment limits and property eligibility for different "Faixas" of the program.
Responsible AI: Grounded the agent's research node using Google Generative AI Embeddings to ensure all technical advice is cited from the processed government manuals.
Gabriel Freire Ushijima is a 19-year-old Computer Science student at USP, São Paulo, passionate about mathematics, optimization, and programming languages. He works across the full stack — proficient in Python, C++, C, Java, HTML, JS, and CSS, with experience in FastAPI, SDL3, numpy, and Matplotlib. His projects range from a research-grade computational geometry solver to a 3D engine built from scratch. Ranked #95 globally on Codewars, he authors problems and supports the community. Open to collaborations, co-founders, and sponsorships.
Computational geometry, mathematical optimization, and autonomous systems. Programming language theory and design. High-performance and systems programming. Competitive programming and algorithm design. Open to collaborating with researchers, engineers, and builders working at the intersection of theory and practical software.
Touring Polygons Problem (IC/USP): Computational geometry optimization using MILP and nonlinear techniques, with autonomous vehicle routing applications. Includes a web visualizer.
3D Engine & Game: Built a full 3D renderer from scratch with SDL3, then developed a game on top of it.
Codewars (#95 global, #36 Python, #1 Brainfuck & Common Lisp): Solve challenges across all skill levels and languages, author problems, and support the community.