Build a platform to onboard applications on on-premise AWS cloud infrastructure. RBAC model for each deployment, Hashicorp Vault for secret management, ssh access to the VMs, PKIs for Kubernetes and compute, all state is code: bitbucket as a database. Jenkins plugin to securely authenticate into vault without sharing secrets. Jenkins plugin to use Vault as a credential store for the whole bank.Golang CLIs provided to the customers.Development of powerful terraform modules for patterned compute deployments.Automatic extraction of the resources in the infrastructure for compliance and monitoring, generation of network diagrams of the deployments using draw.io with node.js.

5th member of the R&D.Contributed to all levels of the product. Mentored new members. Went through 2 pivots of the product.Faced the customers architects for production deployments.Stoic is a Silicon Valley startup that sells a distributed in-memory database for analytics of big data. The visualization tool builds excel like dashboards.• Delivered the packaging and deployment of the software for banks and utility customers: - Deployment on bare-bone Linux servers: non-privileged user, no access to the public internet. - Cluster scheduler with a Service Discovery and Monitoring to controle the cluster: https://nomadproject.io, https://consul.io - Software packaged as a linux container: Docker image. - Runtime for containers: uDocker, a thin alternative to the Docker-engine. Complies with the customers security requirements: no server, no root access, old Linux distros support, no dependencies.• Faced the customers production architects: integration, security and compliance.- Lead the development of the ETL (data-loading) engine. Distributed, concurrent streaming of big-dataset from public stores (AWS S3, AWS Redshift) and private datalakes (Hadoop-HDFS, Hadoop-Hive).• Deployed in pre-production at a private bank: - Time serie: 600M rows processed in parallel by 64 workers per machine. - Small data to enrich the big set: 15 tables 200k-4M rows. - Authentication & Single Sign-On (SAML), logging, monitoring, alerts, source control• Deployment and 3rd level support for all customers.• Developed the text indexer. Identified the data structures and algorithms used by Elasticsearch for fuzzy text queries: Finite State Transducers. Adapted an academical library for language recognition into a text indexer and query executer. C++ and Nodejs.• Facilitated a 30 persons company retreat: drafted a 2 days program with the CEO, team leads. Animated the intro and a 6 Thinking Hats of De Bono session to let the future organization and strategy emerge.

The first iteration of the Stoic product consisted turning a Google App Spreadsheet into a workflow application for SMEs.• Lead the Google Apps Spreadsheets integration.• Lead the deployment on a Platform as a Service (PaaS) - Cloudfoundry.• Successfully automated the deployment of 350 independent servers to fulfil the commitments of a crowd funding campaign. Using Cloudfoundry free hosting with the kind understanding of Pivotal-labs.• Put in place a central monitoring database of the instances to track usage and errors with Elasticsearch.The second iteration of the Stoic product was a development platform for Workflow applications hosted on AWS as a SaaS - Software as a Service.I lead the Elasticsearch integration as a datastore for the object models and raw data:• Layered indices to support multi-tenancy without duplicating data• Versioned indices to support seamless upgrades to the datasets• Elasticsearch as data-lake for big data: Elasticsearch plugin to extract columnar data in parallel data extractions of the columnar data.• seamless migration from a Postgresql traditional database: support both databases in parallel then switch the primary.

I joined Odd-e Scrum and Agile coaching practice in order to train and practice Agile Software Development.• Agile, Scrum, Lean, Facilitation trainings as a student and as a trainer• DevOps and development for Odd-e’s customers and Odd-e’s internals.

Cloud and DevOps.In charge of packaging intalio's software for the cloud.Distribute intalio as a Ubuntu Virtual Machine running on the PaaS Cloud Foundry.Support for AWS/EC2 and private clouds.Transform Intalio's ad-hoc infrastructure into a modern DevOps infrastructure: opscode's Chef, ubuntu's cloud images, debian packaging.Expert Java-OSGi and HTTP server JettyJetty committer and build engineer at Eclipse.org. Early adopter of maven/tycho.Contributed the OSGi integration of jetty on Equinox and Apache/Felix.All of Intalio's java software runs on equinox and Jetty.Migrated Intalio's java software build, tests and internal distribution to OSGi/p2 and maven/tycho.

Lead the re-architecture of Intalio BPMS Designer product from a monolithic swing based application that generates legacy execution code to a modular suite of plugins executing on the Eclipse platform.Architect and main developer of the abstract model for the executable business processes. Architect and lead developer for the data-mapper based on reverse-engineering of XPath2.0 expressions. Working closely with remote co-workers in Ukraine, France and India.

Lead developer of a web application framework that interacts with Business processes and provide rich user applications. Discovered AJAX (XMLHttpRequest) on the browser before the term was invented.

Developed voice activated navigation software in Java.Address capture module written in VoiceXML and executed by the Nuance Recognition Engine.

Prototyped a network of peer-to-peer devices where the number of relayed hops between devices mirrors the physical distance and provide search relevant to the physical environment.This internship was the conclusion of my masters in Computer Science.