independent consulting · cloud data applications

Cloud data applications for mid-sized and large companies

Scoping and delivery of cloud data projects with a broad remit. Most incidents happen at the interfaces: between two clouds, between two teams, between two data regimes. Engagements are short, produce a verifiable deliverable and end with a written handover. They run in French or English, with teams in France or abroad.

The mark draws a switch: two uprights and a moving bar. The bar’s former position stays drawn in a light line, so the previous state and the current one are visible together. The systems delivered follow the same rule. Every change leaves the earlier state visible, and every part can be stopped or replaced without interrupting the others.

01what is on offer

Six ways to engage, from scoping to production. Each ends in a deliverable used in production, which an in-house engineer can read, take over and approve.

build
web apps · analytics · machine learning
Web applications, analytics platforms and machine learning services, from prototype to production on AWS or Azure. Typical duration : 6 months to 1 year.
migrate and modernise
platform migration · legacy code
Cloud platform migration and takeover of existing code. Services that must keep running during the transition stay in service. Typical duration : 3 to 6 months.
audit and govern
system audit · data governance
Critical review of a system already in place, and data governance the team keeps applying after the engagement. Typical duration : 3 months.
hand over and train
new platform · adoption · capability building
Adoption of a new platform by the team that will run it: a use case built with them, the architecture explained, training on their own data. The engagement ends when the in-house team delivers the next dashboard without help. Typical duration : 6 months to 1 year.
adopt AI and change how teams work
generative ai · agents · use cases
Adoption of generative AI and agents: use cases chosen on measured impact, written rules of use, team training. The quality of the result depends mostly on the data given to the model, and what changes is how the business teams work. Typical duration : 6 months to 1 year.
create an online presence
association · liberal profession · independent
Visual identity and online presence for a small organisation. The deliverable includes the design system (tokens, components, written rules) as well as the pages, so the organisation can develop its site without a supplier. This site is one example.
A landscape in sepia ink and wash: a line of trees and low ruins, a cypress at the left, a great deal of paper above.
Claude LorrainRoman Landscape, mid-17th century
02recent engagements

Six real clients, sectors and roles.

2026AI agent strategy for the transport divisionUser support on generative AI, agents built on Copilot, tracking of vendor roadmaps.Egis Group2024Three Azure projects across the industrial businessSand detection in pipelines, refinery cost accounting, on-site risk assessment tool.TotalEnergies — Digital Factory2025HR use case and team training on Microsoft FabricImplementation of the use case, training the in-house teams on the chosen architecture.OCDE
03products

Open-source tools for independent workers, developed outside any client engagement. Their assumptions are shown and can be changed. Two of them below, the others on the products page.

open source · model · ai at work
DUTY

Estimates the time an AI assistant saves over a working day, separating hours, workload and expectations. The gain is capped by review time.

open source · simulator · wealth
FIRE

Estimates your wealth, then what it can pay out each year and how long it lasts, under French or Japanese tax rules.

04latest writing

Articles drawn from the engagements, dated to when the work took place.

2026The grain of randomness: what decides a repetitionA generator of repeated fields is usually controlled by a single number, the quantity of randomness. Yet that is the less important parameter. What the image looks like depends mainly on the level at which the randomness applies, and that level appears nowhere in the code.1 September 20262026Gale-Shapley at the milonga: stable does not mean fairThe algorithm recognised by the 2012 Nobel prize guarantees that no pair of dancers would prefer to form outside the matching produced. It guarantees nobody a dance, and the gap between those two properties is larger than it seems.31 August 2026
A woodcut: geometric solids (an open torus, polyhedra, faceted slabs) set among ruined architecture, rendered entirely in hatching.
Lorenz StöerFantasy of Perspectival Forms Set among Ruins, 1567
05who you work with
Michel Hua
independent consultant · cloud data applications
2010Centrale Lille · Mechanical Engineering · Control Engineering · Applied Mathematics
2018MFG Labs, Havas Media Group
2023Artefact
based inParis · engagements in France and remote
technical stack
distributed computeAthena · Airflow · EMR · Databricks · BigQuery · Functions App · SQL Server · PostgreSQL · OpenAI · Claude · Snowflake · Bedrock · Fabric · Power BI
languagesPython · SQL · Java · Scala · Bash · Terraform
cloudAWS · Azure
certifications
Microsoft Certified Solutions ArchitectMicrosoft Certified AI EngineerMicrosoft Certified Data EngineerMicrosoft Certified Analytics EngineerGoogle Cloud Professional Cloud ArchitectGoogle Cloud Professional Data EngineerSnowflake SnowPro CoreHashiCorp Terraform AssociateGitHub ActionsGitHub Administration