Mateo Amado Ares

Software engineer focused on data pipelines, backend systems, and AI/ML applications.

Experience

Software & AI Engineer, Zepo.ai

Sep 2025 - Present

Build and maintain backend workflows for production use cases across data and operations teams.
Develop and operate Airflow pipelines, Django services, and AI-enabled features for internal decision workflows.
Research and develop new product lines using advances in edge AI.

Software Developer, TriboSolutions

Jan 2025 - Present

Co-founded TriboSolutions and built software for fitness-sector clients, internal tools, and SaaS products.
Build full-stack products end to end — React/TypeScript frontends, backend APIs, and cloud infrastructure.
Implemented CRM integrations and automation while gaining direct client delivery experience.

Machine Learning Engineer, Audialab

Jan 2025 - Feb 2026

Developed AI-powered audio production plugins (VST) with ONNX-based inference.
Moved inference workloads from server-side execution to client-side deployment.

Education

BSc in Computer Engineering, Universidade da Coruña

2021 - 2025

Specialization in computation and AI, with focus on computer vision.
Thesis: Ophthalmological image registration using implicit neural representations.

Selected Projects

Research archives: A Coruña street-image archive (computer vision, OCR search, and map/grid browsing); Spanish Politics on Telegram (searchable archive for research into Spanish politics and Russian influence, updated as messages are collected); El reto de Feijoo (OCR and cross-language search for historical text reuse).
Data mapping: Data centres in Spain — Source-backed map connecting data centres to the electricity transmission grid, with distinct power metrics and visible data gaps. The project attracted interest from a data-centre consulting company.
Production apps: nadus.app (ERP for swimming classes) and eyFisio! (reservation reminder automation).
LLM benchmarks: ee-bench (exploration/exploitation across eight environments), escoba-bench (card-game agent benchmarking), and SciPathBench (citation-path evaluation).
Image competitions: tree_canopy (tree canopy detection with YOLO, EOMT, and Mask2Former), floods_sa (flood mapping in South Africa), and satellite imagery detection.
Hackathons: trace-it (Maisa track winner at HackSpain, team of five; plain-language policies compiled to tested Python with backtesting), VuelingConnect (decentralized communication via BLE), and cMatch (website and Chrome extension to find and manage clothes).
Developer and data tools: beancount-ui (NiceGUI ledger app with banking API integration) and fieldNormalizer (AI-powered merger of heterogeneous data sources); music-stack (self-hosted music discovery and library management, automated tagging, and MCP tools).
Image processing: img-corroder — Rust image glitch tool with SAM2 segmentation.
Fine-tuning: bastos-finetune (text), mosoco (image), and all-the-breaks (audio).

Other Work

Music production, mixing/mastering, and DJ performances in local venues and festivals (YouTube, SoundCloud).
Audiovisual and video work focused on audio-reactive experimentation using Resolume and TouchDesigner.
Co-run No-Cosign, an independent music blog.
Summer camp activity coordination.

Skills

Python, TypeScript, React, FastAPI, Django, Airflow, PostgreSQL, AWS, Go, Data Analysis, AI/ML, Automation, ERP integrations.

Languages

Spanish (native), Galician (native), English (C2)