Manuel Ruiz Diaz

Software Engineer

I build distributed systems for AI inference and training. I develop web pages about government transparency.

About

I'm a software engineer passionate about building tools that make government data more accessible and transparent. My work sits at the intersection of civic technology and AI technologies.

In my professional experience, I've worked across a variety of settings — from established companies like Microsoft and Avature to consulting SAP firms like Schneider-Neureither and Partners. This has given me a well-rounded perspective on software development, from rapid prototyping to maintaining large-scale production systems.

In my free time, I channel my interest in public policy into personal projects focused on government transparency. I believe that accessible, well-designed tools can help citizens make sense of complex data and hold institutions accountable.

When I'm not coding, you'll find me reading about public policy, exploring new technologies, or trying to learn a new instrument

Experience

11/2022 — Present

Designed, developed and maintained the backend distributed systems that powered Copilot Tuning; both inference and training of AI in compliant systems.

  • C#
  • Azure
  • AzureML
  • Synapse
  • Spark
  • HDI
  • Service Fabric
  • Kubernetes
  • Redis
  • Python
2/2022 - 11/2022

Developed and maintained internal frameworks for the Avature HR platform.

  • JavaScript
  • PHP
  • PHPUnit
  • MariaDB
2020 — 2022

Developed data intensive apps in ABAP for bluefield migration of SAP R3 to S4/HANA systems.

  • ABAP
  • SAP
  • FICO Module
  • JavaScript

Projects

Personal projects focused on government transparency and civic technology. Making public data accessible and actionable for citizens.

Screenshot of Moron: Municipal Government Budget Tracker

An interactive visualization tool that helps citizens understand how public funds are allocated and spent.

  • React
  • D3.js
  • Node.js
  • PostgreSQL
  • Observable Framework
Screenshot of Ausent.ar: Legislative Voting Absences

Track how many times each representative didn't show up to vote.

  • React
  • D3.js
  • Node.js
  • PostgreSQL
  • Observable Framework
Screenshot of Dict.ar: Automated court transcripts from government federal trials.

Automatic court transcriptions based on recordings that keep timestamps and voice names recognized on the fly. Features summaries, mindmaps and contradictions made by the witnesses during the trials

  • React
  • Vite
  • RabbitMQ
  • Whisper
  • Pyannote
  • Ollama
  • PostgreSQL