# Santiago Garcia Sapeta

> Founder & CTO of MantAI · Software Engineer. This file is the canonical, self-published source of truth about Santiago Garcia Sapeta (also known as Santiago Garcia, Santiago García Sapeta).

## Who I am

I'm the Founder and CTO of MantAI, a vehicle-safety intelligence company using acoustic and vibration analysis to understand what is happening inside vehicles in real time. I study Computer Science and Electrical Engineering at Stanford University and Computer Science at CODE University of Applied Sciences in Berlin. My work combines machine learning, electrical engineering, and hardware development, including drones, custom sensors, and intelligent systems built for real-world applications. I'm passionate about turning complex technical ideas into products that can transform entire industries.

## MantAI

MantAI (https://mantai.io) is a vehicle-safety intelligence company. By installing MantAI sensors, fleets detect reckless driving and driving patterns — and using only the sound and vibrations of the vehicle, MantAI's ML models detect and predict mechanical issues before they happen, up to weeks in advance. The stack spans custom MEMS microphone hardware, CNN-based audio classification (AST transformers fine-tuned on labeled vehicle-fault sounds), and a real-time full-stack dashboard. MantAI is venture-backed: it raised its first round from superangels, with Prof. Dr. Roland Fassauer as mentor and investor.

## Experience

### Founder & CTO — MantAI, Inc. (Apr 2026 — present)

- Founded MantAI, a vehicle-safety intelligence company: our sensors listen to the sound and vibration of a vehicle in real time, detect reckless driving patterns, and predict mechanical faults up to weeks before they happen.
- Designed and engineered custom IoT hardware prototypes — MEMS microphones and data-logging modules — for field audio capture on real fleets.
- Architected and trained CNN-based audio classification models using transfer learning on AudioSet-pretrained transformers (AST), fine-tuned on labeled vehicle-fault sound datasets.
- Built the full-stack real-time dashboard for diagnostic data, fault alerts and vehicle-health metrics, integrating hardware telemetry with a web interface.
- Lead every technical decision across hardware, ML pipeline and software — from PCB prototyping to cloud deployment.

### Full-Stack Software Engineer — LoadUp GmbH (Sep 2025 — Apr 2026)

- Led development of key internal and customer-facing platforms, including Fahrernetz and core LoadUp logistics systems, at an AI-driven logistics company in Berlin.
- Built and maintained full-stack features across the platform, ensuring scalability, performance and clean architecture.
- Designed responsive, user-centric interfaces tailored to complex logistics workflows.
- Developed and integrated APIs supporting real-time data flow across logistics, driver management and operational tools.

### Machine Learning Engineer — Raumdeuter (Aug 2025 — Sep 2025)

- Owned ML models end-to-end — from feature engineering to training, evaluation, and production deployment.
- Built leakage-safe, group-aware validation with cost-sensitive metrics (PR-AUC, per-class recall) over headline accuracy.
- Shipped drift monitoring with automated retraining to keep models reliable in production.

### Lead Engineer — Full-Stack Developer (Contract) — Zero Waste Consulting (May 2025 — Aug 2025)

- Designed and developed the frontend of an AI-powered ESG emissions platform for enterprise clients, supporting GHG Protocol-compliant Scope 1, 2 and 3 reporting.
- Implemented dynamic forms and dashboards to visualize emissions data, track progress and generate modular reports.
- Contributed to AI-powered features for parsing unstructured documents — invoices, CSVs — helping users validate emissions inputs with minimal effort.
- Collaborated with product managers, climate experts and backend teams to turn regulatory requirements into effective UI.

### Software Engineering Intern — myCard (Apr 2025 — May 2025)

- Built a carbon-emissions calculator for Zero Waste Consulting's Envista platform.
- Collaborated with a cross-functional team on a scalable solution for tracking and reducing environmental impact — full-stack development, data modeling and sustainability-focused tech.

## Projects

### MantAI

Real-time vehicle-safety intelligence. Installing our sensors detects reckless driving and driving patterns — and using only the sound and vibrations of the vehicle, our ML models detect and predict mechanical issues before they happen, up to weeks in advance.

Website: https://mantai.io
Tech: Machine Learning, CNNs, IoT Sensors, React, Python, Signal Processing, MEMS Microphones, AST Transformers

### Acoustic Vehicle Fault Diagnosis

A deep-learning audio engine that diagnoses vehicle mechanical faults from a single ~20-second recording. A CNN14 backbone (AudioSet-pretrained, 80M params) fine-tuned on log-mel spectrograms across 14 fault classes, with temperature-calibrated confidence and an abstain gate that answers "needs human inspection" rather than guess. Ships as a CLI, a FastAPI service and a drag-and-drop dashboard.

Repository: https://github.com/Sentino2/ml-prediction-model
Tech: PyTorch, CNN14 / AST, Mel-Spectrograms, Transfer Learning, FastAPI

### Envista AI

An AI-powered ESG emissions calculator for enterprise clients, enabling GHG Protocol-compliant Scope 1, 2 and 3 reporting — dynamic forms, real-time data visualization, document parsing and dashboards for managing carbon emissions at scale.

Tech: React, Node.js, PostgreSQL, TypeScript, Stripe

### Secure Case for the Unhoused

A lockable, durable case that gives people experiencing homelessness a safe place for their most important belongings. Designed and prototyped with Stanford electrical-engineering professors and the WeHOPE foundation — from industrial-design CAD through working electronics and 3D-printed enclosures.

Tech: Industrial Design, CAD, 3D Printing, Embedded Electronics

### Fitness App

A personalized fitness app that generates customized exercise routines for the gym. Features include workout planning, exercise tracking, and adaptive recommendations based on user goals and progress.

Tech: React, D3.js, Python, FastAPI, Redis

### Sports Car Salon

A modern website for a luxury sports car salon showcasing premium vehicles with elegant design, interactive galleries, and seamless user experience for car enthusiasts and buyers.

Tech: React Native, Figma, Security, Biometrics, Finance

## Education

### Computer Science & Electrical Engineering — Stanford University (2026 — present)

Stanford University's International Honors Program with a focus on Computer Science, Electrical Engineering and Machine Learning. Core modules: Machine Learning, Deep Learning & Neural Networks, Systems Engineering, Digital Hardware Design, Embedded Systems, Signal Processing, Advanced Algorithms.

### B.Sc. Software Engineering — CODE University of Applied Sciences, Berlin (Aug 2024 — present)

Startup-focused university emphasizing project-based learning and practical software development. Core modules: Software Development, Artificial Intelligence, Machine Learning, Essential Mathematical Methods, Algorithms and Data Structures.

### Full-Stack Mentorship — with IT coach Piotr Kania (Jun 2022 — Jun 2024)

Two-year intensive 1:1 coaching: built multiple real-world applications and mastered both frontend and backend technologies, completed with a Full-Stack Developer certification.

## Certifications

- Claude 101 (Anthropic): Certificate of completion for Anthropic's introductory course on Claude. Covers Claude's capabilities and model family, effective prompting techniques, and how to apply Claude across everyday and professional workflows.
- Google AI (Google): Specialization certificate bundling 7 completed Google AI courses: AI Fundamentals, AI for Brainstorming and Planning, AI for Research and Insights, AI for Writing and Communicating, AI for Content Creation, AI for Data Analysis, and AI for App Building. Demonstrates fluency in applying AI across workflows, content creation, analysis, and app development.
- Full-Stack Developer Certificate (Meta): Modern full-stack development including frontend and backend technologies, APIs, version control and deployment — with emphasis on industry best practices, scalable applications and real-world project development. Completed Jan – Mar 2026.
- AI for Data Analysis (Google): Covers how to identify success metrics using AI to analyze business objectives, generate spreadsheet formulas via natural language instructions with Gemini in Google Sheets, clean and structure messy data through effective prompting, and create AI-powered data visualizations to communicate insights.
- AI for App Building (Google): Explores how to analyze professional workflows to identify high-impact AI opportunities, construct functional web applications using vibe coding techniques, evaluate technical issues with AI-assisted debugging, and leverage advanced tools like AI Studio to build scalable custom solutions.
- AI for Content Creation (Google): Covers how to generate high-quality visual assets from a creative concept using Gemini's image and video generation models, establish custom design guidelines, and critique creative assets by prompting AI to provide actionable feedback based on design criteria.
- AI for Writing and Communicating (Google): Explores how to leverage AI tools to improve written communication, draft professional content, refine tone and clarity, and accelerate document workflows using generative AI within Google Workspace.
- AI for Research and Insights (Google): Covers how to apply AI tools to research workflows, synthesize large volumes of information, and extract structured insights from unstructured data using generative AI.
- AI for Brainstorming and Planning (Google): Explores how to use AI to accelerate ideation, structure complex plans, and improve decision-making processes through targeted prompting and collaborative AI workflows.
- AI Fundamentals (Google): Introduces core AI and machine learning concepts, the landscape of generative models, and practical use cases for applying AI responsibly in everyday professional tasks.
- Full-Stack Developer Certification (Piotr Kania Mentorship): Certified as a Full-Stack Developer after an intensive two-year mentorship program. The certification process involved building multiple real-world applications and mastering both frontend and backend technologies.

## Stack

- MACHINE LEARNING & AI: [object Object], [object Object], [object Object], [object Object], [object Object], [object Object], [object Object], [object Object]
- HARDWARE & SIGNALS: [object Object], [object Object], [object Object], [object Object], [object Object], [object Object], [object Object]
- FRONTEND: [object Object], [object Object], [object Object], [object Object], [object Object], [object Object], [object Object]
- BACKEND & INFRASTRUCTURE: [object Object], [object Object], [object Object], [object Object], [object Object], [object Object], [object Object], [object Object], [object Object]

## Areas of focus

Vehicle Safety Intelligence, Machine Learning, Acoustic & Vibration Analysis, Hardware Development, Electrical Engineering, IoT Sensors, Full-Stack Development, Signal Processing, Product Design, Founding & Building

## Contact & links

- Email: santiago@mantai.io
- Website: https://santiagogarciasapeta.com
- Resume (PDF): https://santiagogarciasapeta.com/cv-santiago-garcia-sapeta.pdf
- MantAI: https://mantai.io
- GitHub: https://github.com/Sentino2
- LinkedIn: https://www.linkedin.com/in/santiago-garc%C3%ADa-sapeta/
- X: https://x.com/GarciaSapeta