César de Paula Morais

Hotmart

Creator-economy platform

Software Engineer

Sep 2025 – Present

Software Engineer InternConverted to full-time ahead of schedule

Oct 2024 – Aug 2025

• —Building Launchpad, Hotmart's flagship prompt-to-app AI builder, across front end, back end, and multi-agent architecture (LangGraph, tool calling, MCP, guardrails, evals) — the only non-staff engineer on the team.
• —Co-built, from a hackathon, an agent platform that turns product specs into production-ready PRs via tool calling, MCP, and evals; as main owner, drove it to 1,000+ merged PRs and a company-wide migration, aligning teams and leadership.
• —Built from scratch an agentic platform to ship AI-generated apps to production with security and compliance guardrails: 50+ apps live, setup cut from days to minutes.
• —Co-owned Hotmart's Internal Developer Platform (Backstage, Node.js, React) for 800+ engineers across 50+ teams: shipped full-stack features including RAG-based knowledge search, supported users, reviewed 140+ pull requests, and introduced agile rituals the team still runs.
• —Technical lead for Launchpad content and community, working directly with founders (meetup with 60+ founders and builders); mentored 3 interns who now lead their own projects.

Python / TypeScript / React / AWS / LangGraph / MCP / Evals / RAG / Backstage

Adimplere

Debt collection platform

Backend Developer

Sep 2022 – Sep 2023

• —Built and integrated backend microservices for billing systems, with automated testing and Docker-based deployments on AWS.

Python / FastAPI / Django / Pytest / Docker / AWS

UFMG

Universidade Federal de Minas Gerais

B.Sc. in Computer Science

2021 – Jun 2026

• —Coursework: Data Structures, Algorithms, Operating Systems, Computer Networks, Software Engineering.

UFMG

Undergraduate thesis

Adaptive Bitrate for Cloud Gaming

2026

Advisor: Prof. Luiz Filipe Menezes Vieira

• —Built MACCA, a server-side adaptive bitrate controller for Sunshine (open-source cloud gaming server), adapting TCP CUBIC's congestion control to a 1 Hz, RTT-only loop with no protocol or client changes.
• —Cut peak RTT by up to 43% and stall time by up to 61% in 5 of 6 emulated network scenarios on real hardware; accepted for oral presentation at SBESC 2026.

C++ / GoogleTest / Networking

WiNET · UFMG

Wireless Networks Lab

Undergraduate Research Assistant

Sep 2023 – Sep 2024

• —Built supervised, unsupervised, and neural-network models to detect attacks on 5G core networks (NWDAF), using dimensionality reduction to improve accuracy and efficiency; published at GIIS 2025.

Python / scikit-learn / pandas