Experience

Owkin

• United States
• Biotechnology Research
• 200 - 300 Employee

• Software Engineer

  • Jan 2022 - Present

  Work on Substra, an open source federated learning framework designed with medical research in mind. I focus particularly on deployment on Kubernetes via Helm, possibly involving Terraform; I've worked on GCP, on-premises, and also assisted external partners for their own deployments. I've also taken up most work on the internal CI, which uses GitHub Actions and some amount of Python code. I've also worked on Substra itself, particularly the Django backend, but at times even the… Show more Work on Substra, an open source federated learning framework designed with medical research in mind. I focus particularly on deployment on Kubernetes via Helm, possibly involving Terraform; I've worked on GCP, on-premises, and also assisted external partners for their own deployments. I've also taken up most work on the internal CI, which uses GitHub Actions and some amount of Python code. I've also worked on Substra itself, particularly the Django backend, but at times even the frontend (React, JS) and components in Go.

• DevOps Engineer

  • Apr 2019 - Jan 2022

  Primary role is improving the CI & CD infrastructure for the development of Owkin Studio. Most notably: - Porting Owkin Studio from Docker-compose to Kubernetes - Then porting Studio from homemade Kubernetes templates to a Helm chart. - Working on our software factory and Jenkins (who could have thought Groovy was so good?) - Working on setting up (EFK-based) reporting (who could have thought ES queries were so bad?) - Managing web traffic (most notably but not… Show more Primary role is improving the CI & CD infrastructure for the development of Owkin Studio. Most notably: - Porting Owkin Studio from Docker-compose to Kubernetes - Then porting Studio from homemade Kubernetes templates to a Helm chart. - Working on our software factory and Jenkins (who could have thought Groovy was so good?) - Working on setting up (EFK-based) reporting (who could have thought ES queries were so bad?) - Managing web traffic (most notably but not exclusively through Traefik) on deployed instances - Making a web interface for the software factory, in Django. - Making some tools in Python for developers - Making some tools in unholy combinations of bash/awk/groovy/dlang for myself - Taking revenge on the world for making me read so much documentation by writing some more myself Plus, as Owkin still is a human-sized company, tasks elsewhere where I can be useful. Unplugging network cables, repairing MacBooks, setting up Pypi servers.. I aim to please.



ASML

• Netherlands
• Semiconductor Manufacturing
• 700 & Above Employee

• Intern (Test infrastructure)

  • Jun 2018 - Nov 2018

  ASML's core business is the design and manufacture of photolithography equipment — that is, they make machines that project circuit layouts onto silicon for mass production. ASML has been investing in a new technology, extreme ultraviolet (EUV), for a decade. EUV machines are cutting edge, being kept under constant vacuum to preserve the near perfect curve (± 2 atoms) of the mirrors and avoid absorbing the 13.5nm light, the production of which requires laser vaporization of tin droplets… Show more ASML's core business is the design and manufacture of photolithography equipment — that is, they make machines that project circuit layouts onto silicon for mass production. ASML has been investing in a new technology, extreme ultraviolet (EUV), for a decade. EUV machines are cutting edge, being kept under constant vacuum to preserve the near perfect curve (± 2 atoms) of the mirrors and avoid absorbing the 13.5nm light, the production of which requires laser vaporization of tin droplets into plasma to produce the light. ASML only sells a very approximate 200 new machines a year. It would not be cost-effective to test all software onto actual machines. ASML thus uses solutions simulating machines to various degrees of accuracy (generally inversely proportional to cost or availability). My task was to improve the reliability of some of these machines. Certainly the most interesting task so far has been the remote back-up of machines. I researched boot mechanisms to implement a custom boot sequence where a kernel and a custom initial ramdisk were transferred and loaded into RAM, then mounted the disk with the data to save, and rsynced it to the remote server before rebooting the machine; thus operating without ever installing anything onto the target machines. The point of this was to improve the availability of the test infrastructure : restoring from heavy problems (such as a machine no longer properly booting) is now a matter of hours, and machines can be kept consistent and free of corruption at little cost. Show less ASML's core business is the design and manufacture of photolithography equipment — that is, they make machines that project circuit layouts onto silicon for mass production. ASML has been investing in a new technology, extreme ultraviolet (EUV), for a decade. EUV machines are cutting edge, being kept under constant vacuum to preserve the near perfect curve (± 2 atoms) of the mirrors and avoid absorbing the 13.5nm light, the production of which requires laser vaporization of tin droplets… Show more ASML's core business is the design and manufacture of photolithography equipment — that is, they make machines that project circuit layouts onto silicon for mass production. ASML has been investing in a new technology, extreme ultraviolet (EUV), for a decade. EUV machines are cutting edge, being kept under constant vacuum to preserve the near perfect curve (± 2 atoms) of the mirrors and avoid absorbing the 13.5nm light, the production of which requires laser vaporization of tin droplets into plasma to produce the light. ASML only sells a very approximate 200 new machines a year. It would not be cost-effective to test all software onto actual machines. ASML thus uses solutions simulating machines to various degrees of accuracy (generally inversely proportional to cost or availability). My task was to improve the reliability of some of these machines. Certainly the most interesting task so far has been the remote back-up of machines. I researched boot mechanisms to implement a custom boot sequence where a kernel and a custom initial ramdisk were transferred and loaded into RAM, then mounted the disk with the data to save, and rsynced it to the remote server before rebooting the machine; thus operating without ever installing anything onto the target machines. The point of this was to improve the availability of the test infrastructure : restoring from heavy problems (such as a machine no longer properly booting) is now a matter of hours, and machines can be kept consistent and free of corruption at little cost. Show less



DxO Labs

• France
• Photography
• 1 - 100 Employee

• Intern (Test infrastructure)

  • Jun 2017 - Aug 2017

  DxO is an image processing company, which released its own camera, the DxO ONE, in 2015. Its novel concept involves connecting with a smartphone, through an application. I got to work on the camera's automated test suite, distributed across a network of cameras connected to dedicated computers in a controlled environment. Most notably, I worked on implementing a new way of distributing tests through the HTCondor set of distributed computing tools. The test array can be thought of… Show more DxO is an image processing company, which released its own camera, the DxO ONE, in 2015. Its novel concept involves connecting with a smartphone, through an application. I got to work on the camera's automated test suite, distributed across a network of cameras connected to dedicated computers in a controlled environment. Most notably, I worked on implementing a new way of distributing tests through the HTCondor set of distributed computing tools. The test array can be thought of as a cluster of machines, each with their own capabilities : we distribute jobs (tests) based on capabilities. This allowed for greater reliability (and resilience in case of crashes), better reproductibility of tests, and much improved scalability. In addition to HTCondor, I also got the opportunity of practicing my Python skills (on existing infrastructure) and learned about Ansible (allowing to run commands across a network). I also got to discover the tools used at DxO for organization : common software products, but also Large-scale Scrum (LeSS), enabling the all-around great team I got to be a part of to work in unisson. Show less DxO is an image processing company, which released its own camera, the DxO ONE, in 2015. Its novel concept involves connecting with a smartphone, through an application. I got to work on the camera's automated test suite, distributed across a network of cameras connected to dedicated computers in a controlled environment. Most notably, I worked on implementing a new way of distributing tests through the HTCondor set of distributed computing tools. The test array can be thought of… Show more DxO is an image processing company, which released its own camera, the DxO ONE, in 2015. Its novel concept involves connecting with a smartphone, through an application. I got to work on the camera's automated test suite, distributed across a network of cameras connected to dedicated computers in a controlled environment. Most notably, I worked on implementing a new way of distributing tests through the HTCondor set of distributed computing tools. The test array can be thought of as a cluster of machines, each with their own capabilities : we distribute jobs (tests) based on capabilities. This allowed for greater reliability (and resilience in case of crashes), better reproductibility of tests, and much improved scalability. In addition to HTCondor, I also got the opportunity of practicing my Python skills (on existing infrastructure) and learned about Ansible (allowing to run commands across a network). I also got to discover the tools used at DxO for organization : common software products, but also Large-scale Scrum (LeSS), enabling the all-around great team I got to be a part of to work in unisson. Show less



Materialise 3D

• France
• Technology, Information and Internet
• 1 - 100 Employee

• Intern (Web development)

  • Jul 2016 - Jul 2016

  Materialise 3D is (or, at the time, was) a two-people start-up. Born from the Materializacion 3D foundation (providing 3d-printed prostheses in Colombia), it sought at the time to overhaul its directory of "makers", Red3D (http://www.red3d.org/). "Makers" are essentially the hackers of the "do-it-yourself" culture. Red3D's web stack notably included WordPress, Bootstrap, and a set of proprietary scripts. I produced a new Wordpress theme (integrating commands for the rest of the stack… Show more Materialise 3D is (or, at the time, was) a two-people start-up. Born from the Materializacion 3D foundation (providing 3d-printed prostheses in Colombia), it sought at the time to overhaul its directory of "makers", Red3D (http://www.red3d.org/). "Makers" are essentially the hackers of the "do-it-yourself" culture. Red3D's web stack notably included WordPress, Bootstrap, and a set of proprietary scripts. I produced a new Wordpress theme (integrating commands for the rest of the stack, which required understanding of it) to fulfill the new requirements for the website. I also cleaned up an unfortunate duplication incident on the (SQL) database with no loss of data. Unfortunately, I only got to spend one month working with the fine people at M3D before the end of my internship. Show less Materialise 3D is (or, at the time, was) a two-people start-up. Born from the Materializacion 3D foundation (providing 3d-printed prostheses in Colombia), it sought at the time to overhaul its directory of "makers", Red3D (http://www.red3d.org/). "Makers" are essentially the hackers of the "do-it-yourself" culture. Red3D's web stack notably included WordPress, Bootstrap, and a set of proprietary scripts. I produced a new Wordpress theme (integrating commands for the rest of the stack… Show more Materialise 3D is (or, at the time, was) a two-people start-up. Born from the Materializacion 3D foundation (providing 3d-printed prostheses in Colombia), it sought at the time to overhaul its directory of "makers", Red3D (http://www.red3d.org/). "Makers" are essentially the hackers of the "do-it-yourself" culture. Red3D's web stack notably included WordPress, Bootstrap, and a set of proprietary scripts. I produced a new Wordpress theme (integrating commands for the rest of the stack, which required understanding of it) to fulfill the new requirements for the website. I also cleaned up an unfortunate duplication incident on the (SQL) database with no loss of data. Unfortunately, I only got to spend one month working with the fine people at M3D before the end of my internship. Show less



Thales

• France
• IT Services and IT Consulting
• 700 & Above Employee

• Intern

  • May 2015 - Jul 2015

  I worked at Thales Communications & Security, on the TRC-6500 strategic high-frequency sensor. I researched and implemented a solution to automate the installation of a customized Debian OS for a computing cluster, working largely in autonomy as to the specifics of the solution, while receiving all the necessary help and input to progress from my supervisor. This internship enabled me to learn much about Debian specifics, such as the installer and the package manager. Our solution… Show more I worked at Thales Communications & Security, on the TRC-6500 strategic high-frequency sensor. I researched and implemented a solution to automate the installation of a customized Debian OS for a computing cluster, working largely in autonomy as to the specifics of the solution, while receiving all the necessary help and input to progress from my supervisor. This internship enabled me to learn much about Debian specifics, such as the installer and the package manager. Our solution also made me get familiar with SimpleCDD, itself a Perl script using DebianCD. Finally, I got much more comfortable with standard UNIX tools. Show less I worked at Thales Communications & Security, on the TRC-6500 strategic high-frequency sensor. I researched and implemented a solution to automate the installation of a customized Debian OS for a computing cluster, working largely in autonomy as to the specifics of the solution, while receiving all the necessary help and input to progress from my supervisor. This internship enabled me to learn much about Debian specifics, such as the installer and the package manager. Our solution… Show more I worked at Thales Communications & Security, on the TRC-6500 strategic high-frequency sensor. I researched and implemented a solution to automate the installation of a customized Debian OS for a computing cluster, working largely in autonomy as to the specifics of the solution, while receiving all the necessary help and input to progress from my supervisor. This internship enabled me to learn much about Debian specifics, such as the installer and the package manager. Our solution also made me get familiar with SimpleCDD, itself a Perl script using DebianCD. Finally, I got much more comfortable with standard UNIX tools. Show less