Mario Dagrada

Mario Dagrada

Engineering manager for scientific software

Hello everyone!

I am Mario, an engineering manager with a strong technical background in scientific software development. I am currently based in Eindhoven, the Netherlands.

I use this website mainly for tracking my professional career and showcase some of the projects I have been working on. I also occasionally (and hopefully in the future more regularly) write blog posts on scientific software engineering, cloud computing, DevOps, and machine learning.

Interests

We are drowning in information but starved for knowledge.
– John Naisbitt

I always loved working in small companies where my work has an immediate (and meaningful) impact. Currently, my main professional interests are:

  • Scientific software development: this combines my two main passions of programming and science, especially physics. I have experience in HPC, cybersecurity, and quantum computing.
  • Engineering and product management: I love to start, grow, and continuously improve engineering teams and do the same the with software products, especially if they involve scientific/engineering foundations.
  • Cloud infrastructure and distributed applications: I have experience in building distributed systems running on public cloud, always using infrastructure-as-a-code with no compromises.
  • Machine learning for science and AI engineering: ML is a huge field. I have extensive experience with scientific ML, time series forecasting, and recently I ventured into the field of AI engineering, building AI infrastructure leveraging large language models.

You can find more details on my previous works in the blog and projects or my Google Scholar profile.

Career

Per aspera ad astra.
– The city of Gouda (among others)

Detailed CV in PDF

  • Jul 2024 - Present - AI engineer and consultant (freelance): I put into practical use my expertise in software architecture and development by helping several companies in building their AI infrastructure and overall strategy. A couple of project highlights:

    • Jupi: I built the whole AI infrastructure from scratch for Jupi, a cool SaaS product for taking decentralized decision within companies. I started very early in the company journey, setting up the LLM call pipelines, creating tools for evaluating the quality of the prompts and perform prompt engineering, adding tools and agentic capabilities to the system, and (still ongoing), adding the whole memory layer to the AI agents for giving more informed answers to the decision-making.

    • Onderlinge: I made a 6-months PoC for this established insurance company in the Netherlands. The goal was to leverage AI agents to analyze and produce automatically the documentation of a really large legacy codebase, written in multiple languages, and very poorly documented. The finaly accomplishment has been the automatic creation of a knowledge graph useful for navigating the codebase and supported future migration initiatives in the company.

  • Feb 2024 - Jul 2025 - CTO at Friday Energy (scientific software, renewable energies): I led and contributed to the modernization of the whole software platform (migrating from Python to Golang), doubled the size of the software team, started and nurtured the collaboration with several key partners, and strongly contributed to the roadmap and strategic company direction in terms of software and algorithm decisions. I could not do that without the help of several brilliant and highly motivated colleagues in the team.

  • Sep 2020 - Feb 2024 - Head of quantum software at PASQAL (scientific software, quantum computing): I planned and executed the delivery of the first company cloud-based product. After a promotion (and an acquisition), I managed and grew up to 10 members a team of scientific software developers working on cutting-edge libraries for quantum computing and scientific machine learning.

  • Aug 2023 - AWS Certified Solutions Architect, Associate: AWS certification testing the knoweldge of its cloud services for building high available and secure applications.

  • Apr 2019 - Jun 2019 - Design-Driven Business Innovation at Amsterdam Business School: an intensive 3-months course which taught me how to ideate, prototype and execute innovation projects within a corporate environment. The course ended with an innovation project proposal which was accepted at my company.

  • Apr 2018 - Sep 2020 - Senior R&D software engineer and innovation lead at Forescout (cybersecurity): a lot of coding and machine learning algorithms to improve the life of security analysts. After a promotion, I became the head of the product innovation team of the company where I led the prototype and later integration into the main product line of several projects.

  • Aug 2016- Mar 2018 - R&D software engineer at Atos Bull (scientific software, HPC): a lot of low-level coding to make the interconnect of Atos supercomputers faster and help scientists running their codes at maximum efficiency. This has also been my first encounter with Agile since I acted as Scrum Master of a team of 8 for over a year.

  • Oct 2013- Oct 2016 - PhD in computational physics at Sorbonne University in Paris: simulations on high-temperature superconductors and a lot of writing, both scientific papers and a long thesis but especially highly efficient parallel scientific software.

  • Sep 2011- Sep 2013 - Master in Physics of Complex Systems from the Polytechnic University of Turin with a thesis on material simulations for fuel cells carried out in Paris.

Some of my favourites

The passion for technology has always been at a core of my work, and I have a few ones I especially enjoyed working with:

  • PyTorch: by far my favorite deep learning library due to its very intuitive and Pythonic interface, nice and flexible differentiation engine, and super helpful community. It is losing some traction nowadays, maybe at some point I will switch to Jax.

  • Vercel AI SDK: the best tool for AI engineering if using Typescript. The sheer amount of features and use cases supported can be intimidating, but once mastered it, one can basically do everything with AI model at a very high pace.

  • FastAPI: the best library to build backend APIs, in my opinion. And the quality of its documentation is something rarely found in open-source projets.

  • SKtime: provides a unified interface for time series analysis with excellent scikit-learn integration and great extensibility. There is basically everything for time-series in here while still managing to keep a decent interface for the users.

  • Golang: the language has its downsides (so hard to build proper tests, for example), but its concurrency programming model is, in my opinion, unparalleled and really makes the difference for large scale backend applications. And well, it powers Kubernetes, one of the largest and most contributed open-source software in the world.

  • Pulumi: enables infrastructure-as-code using familiar programming languages and a super nice developer experience both in the cloud and with the CLI. One needs to be particularly careful to properly organize the code, but then, unlike Terraform, it becomes a real pleasure to work with.

  • PostgreSQL: no need for explanations here, 40 years old and still going strong with really good features included at each new major release. Actually, later it is doing better than most other NoSQL competing solutions, and it is becoming more than just a database engine.