Mateus N.

𝗦𝘁𝗼𝗽 𝗵𝗶𝗿𝗶𝗻𝗴 𝗳𝗼𝗿 "𝗳𝗲𝗮𝘁𝘂𝗿𝗲𝘀." 𝗛𝗶𝗿𝗲 𝗳𝗼𝗿 𝗼𝘂𝘁𝗰𝗼𝗺𝗲𝘀. 𝗜 𝗯𝘂𝗶𝗹𝗱 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻-𝗴𝗿𝗮𝗱𝗲, 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝘁𝗵𝗮𝘁 𝘀𝗰𝗮𝗹𝗲. I specialize in White-Collar Automation: transforming complex, high-stakes human workflows into high-margin AI engines that move the needle for your business. PROVEN TRACK RECORD - From $0 to $225k/mo in 10 Months: As Founder and CTO at Jumpstart, I built the AI legal engine that manages US immigration petitions end-to-end. I replaced manual labor with high-precision autonomous agents that handle high-stakes legal logic. - The "Sequoia & Founders Fund" Filter: As Founding Engineer at ENTER (Enterprise AI), I helped scale the company to a $350M valuation with $40M+ in funding from Sequoia Capital and Founders Fund. I personally engineered the AI solutions used to close enterprise giants like Nubank (publicly traded, $80B+ market cap). - ML at Scale for a $3B Unicorn: As Data Scientist at Wildlife Studios (backed by Benchmark), I built ML pipelines processing terabytes of data and delivered a pricing model that generated +15% expected profit (~$45k/month additional) while maintaining ROAS. PREMIUM ACADEMIC BACKGROUND - Double ITA Alum: I hold both a Bachelor's and a Master's in Computer Engineering from ITA (the "Brazilian MIT"), specializing in Large Language Models (LLMs). - Elite Problem Solver: Multi-medalist in National Math, Physics, and Chemistry Olympiads. I bring a rigorous, first-principles approach to architectural challenges that require complex logical reasoning. TECH STACK - AI & Multi-Agent Orchestration: 𝗔𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀 𝗮𝗴𝗲𝗻𝘁𝗶𝗰 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 and RAG pipelines using 𝗣𝘆𝗱𝗮𝗻𝘁𝗶𝗰 𝗔𝗜, 𝗟𝗮𝗻𝗴𝗚𝗿𝗮𝗽𝗵, 𝗟𝗹𝗮𝗺𝗮𝗜𝗻𝗱𝗲𝘅, and CrewAI. 𝗗𝘂𝗿𝗮𝗯𝗹𝗲 𝗲𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 at scale with 𝗛𝗮𝘁𝗰𝗵𝗲𝘁, 𝗧𝗲𝗺𝗽𝗼𝗿𝗮𝗹, and Inngest. Multi-modal integration with 𝗢𝗽𝗲𝗻𝗔𝗜 (GPT-5), 𝗖𝗹𝗮𝘂𝗱𝗲 (4.5 Sonnet/Haiku), 𝗚𝗲𝗺𝗶𝗻𝗶 (3 Flash), and DeepSeek. Advanced retrieval: 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗥𝗔𝗚, 𝗚𝗿𝗮𝗽𝗵𝗥𝗔𝗚, and Hybrid Search (BM25 + vector). Robust 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻 𝗵𝗮𝗿𝗻𝗲𝘀𝘀𝗲𝘀 (evals) to ensure reliability and minimize hallucinations in production. - Voice, Video & Generative Media: Voice agents with 𝗩𝗮𝗽𝗶, 𝗥𝗲𝘁𝗲𝗹𝗹 𝗔𝗜, and Bland AI. Speech processing with 𝗚𝗿𝗼𝗾 𝗪𝗵𝗶𝘀𝗽𝗲𝗿, 𝗗𝗲𝗲𝗽𝗴𝗿𝗮𝗺, 𝗘𝗹𝗲𝘃𝗲𝗻𝗟𝗮𝗯𝘀, and OpenAI TTS. Video generation with 𝗛𝗲𝘆𝗚𝗲𝗻, Runway, Pika Labs, and Synthesia. Image generation with 𝗜𝗺𝗮𝗴𝗲𝗻 𝟰, 𝗗𝗔𝗟𝗟-𝗘 𝟯, Flux, Midjourney, and Stable Diffusion. - Document Processing & Data Extraction: Intelligent document parsing with 𝗗𝗼𝗰𝗹𝗶𝗻𝗴, 𝗨𝗻𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲𝗱, 𝗟𝗹𝗮𝗺𝗮𝗣𝗮𝗿𝘀𝗲, Amazon Textract, and Google Document AI. LLM-ready web scraping with 𝗙𝗶𝗿𝗲𝗰𝗿𝗮𝘄𝗹, 𝗖𝗿𝗮𝘄𝗹𝟰𝗔𝗜, Browserbase, ScrapeGraphAI, and Apify. Chatbot platforms: 𝗩𝗼𝗶𝗰𝗲𝗳𝗹𝗼𝘄, Botpress, Intercom Fin, and Chainlit. - Full-Stack Engineering: High-performance web apps with 𝗡𝗲𝘅𝘁.𝗷𝘀, 𝗥𝗲𝗮𝗰𝘁, and shadcn/ui. Scalable backends using 𝗣𝘆𝘁𝗵𝗼𝗻 (𝗙𝗮𝘀𝘁𝗔𝗣𝗜) and 𝗡𝗼𝗱𝗲.𝗷𝘀. Rapid UI prototyping via 𝗟𝗼𝘃𝗮𝗯𝗹𝗲 and 𝘃𝟬. - No-Code & Low-Code Automation: 𝗻𝟴𝗻 (self-hosted), 𝗠𝗮𝗸𝗲, 𝗭𝗮𝗽𝗶𝗲𝗿, and Windmill for business workflow automation and AI integration. - AI-Augmented Development: Full fluency in 𝗖𝘂𝗿𝘀𝗼𝗿, 𝗖𝗹𝗮𝘂𝗱𝗲 𝗖𝗼𝗱𝗲, 𝗟𝗼𝘃𝗮𝗯𝗹𝗲, and 𝗢𝗽𝗲𝗻𝗖𝗹𝗮𝘄 as daily workflow tools for faster iteration, less boilerplate, and higher output. - Data Engineering: Large-scale pipelines with 𝗗𝗮𝘁𝗮𝗯𝗿𝗶𝗰𝗸𝘀, 𝗣𝘆𝗦𝗽𝗮𝗿𝗸, and 𝗔𝗶𝗿𝗳𝗹𝗼𝘄. ETL/ELT workflows, data warehousing, A/B testing frameworks, and ML training/deployment pipelines. - Security & Compliance: 𝗦𝗦𝗢/𝗢𝗜𝗗𝗖/𝗢𝗔𝘂𝘁𝗵𝟮, 𝗥𝗕𝗔𝗖, audit logging, and secrets management. Systems designed for 𝗦𝗢𝗖 𝟮, 𝗛𝗜𝗣𝗔𝗔, 𝗚𝗗𝗣𝗥, and 𝗣𝗖𝗜 𝗗𝗦𝗦 (hands-on experience at Jumpstart). - Cloud & DevOps: Production-ready infrastructure on 𝗔𝗪𝗦, 𝗚𝗖𝗣, and 𝗔𝘇𝘂𝗿𝗲. Environment provisioning via 𝗧𝗲𝗿𝗿𝗮𝗳𝗼𝗿𝗺 (IaC), containerization with 𝗗𝗼𝗰𝗸𝗲𝗿 and 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀, and monitoring with Sentry, Grafana, and Prometheus. WHY MOST AI PROJECTS FAIL (AND WHY MINE DO NOT) I operate as your Technical Partner, not just a dev. 1. Weekly Loom Videos: Every Friday, you receive a video walkthrough of exactly what was built and what is coming next. No guessing, no ghosting. 2. Notion Project HQ: 24/7 access to a central "Source of Truth" containing all project requirements, specs, and a history of every update. 3. No-BS Consulting: I provide honest advice in your interest, even if it means telling you a project is not worth building. THE METHOD: SKATEBOARD TO SUPERCAR - The Skateboard: A functional, "thin-slice" MVP in 2 weeks to prove core value. - The Motorcycle: Scaling orchestration, hardening security, and adding robust evals. - The Supercar: A high-performance, autonomous system with 99.9% reliability.