Everything as Code
When a business rule or an architecture decision matters, I write it down in Git.
Teams find the context there, and so do AI agents.
IT transformation, product & AI
I speed up your products without weakening your IT.
Architect and transformation CTO. I step in when the existing system slows the product down or costs too much to change.
My test stays simple. What we change has to cut a risk or an operating cost.
- 15 years
- transforming IT systems
- 8 sectors
- from optics to finance
- ~100 projects
- industrialised in CI/CD
- 14 applications
- shipped from brief to production
Companies I've worked with
- Carrefour
- Essilor
- BPCE
- CNRS Innovation
- Saretec
- Dessange International
- Stannah
AI as a way of working
AI that helps once the context is solid
I use AI every day. It gives good results when the context is solid. The real work is giving it the business and its constraints.
From there, some projects finally become reasonable.
Daily use, kept in check
I use it from scoping to tests, never on autopilot.
What it writes meets the same bar as the rest of the code: SonarQube static analysis, dependency checks, human review.
Training the teams
I train teams on their actual work.
They should come out self-sufficient, with practices they can keep.
What I do
Areas I work in
I can come in early during scoping or go all the way down to the code and the release. Either way, I keep the trade-off at the centre. Operating cost, risks on the ground.
AI strategy & internal platform
I start with the uses where the effort pays off.
Then I put in place the base and the quality rules that keep prototypes from being forgotten.
AI applied to the business
Compliance review, document extraction.
I build these features with the business teams, then harden them through to production.
Software design & DDD
Software should use the words of the business.
I clarify the domain and its boundaries.
That is also what gives AI agents something solid to hold on to.
IT & cloud modernisation
I start with the foundations.
We move in steps short enough to hold, with a clear trade-off on cost and operations.
Clean-up, FinOps & security
I look for what still costs money while no longer earning its keep.
Forgotten servers, dormant subscriptions.
We remove, then we measure.
Data, DevOps & industrialisation
ETL, SQL and APIs depending on the context.
I tidy up the data flows and make them observable.
On the delivery side, deployments move out of manual steps into repeatable pipelines.
Measured results
Effects measured on recent cases
The cases are anonymised. The orders of magnitude stay precise enough to discuss with their context in a call.
−€200k/yrin recurring costs
Recurring costs that had settled in over the years
Some tasks were handed to outside providers.
Others sat idle inside subscriptions.
Once the teams and the AI agents were equipped, the work came back in-house.
The cloud bill dropped by 15% and old servers moved to managed services.
−90%on maintenance cost
A custom-built site that was expensive to maintain
It tied up two developers and created a constant technical dependency.
I recommended a managed solution despite the early reservations.
Maintenance no longer ties up the team.
Marketing took back control of its own content.
2 monthsfrom design to production
A control tower for an entire real-estate portfolio
An internal application that brings data scattered across three business systems and around fifteen external sources into a single view.
It tracks profitability, vacancy, unpaid rent and ESG indicators.
Leases and works commitments left Excel for the tool.
< 30 minto restore the service
An overloaded server during tax season
The overload threatened to bring the service down at peak load.
I diagnosed and restored it before users were affected, with help from AI.
A typical engagement
A transformation, step by step
I modernise without gambling with production. The work starts with what blocks or exposes the most, then moves forward until the team can carry on without me.
Talk through your needUnderstand
I start from the business and from where it hurts.
I check early on the ground, then version what we decide as shared memory.
Stabilise
I secure the foundations and the access paths.
Obsolete components are updated or removed from the system.
Simplify
I cut the debt and what it costs every month.
A shared base replaces the one-off setups.
Modernise
I modernise when the benefit is clear.
Product owners and developers ship faster with coding agents.
Developers keep watch over the architecture and the hard problems.
Track record
Fifteen years of delivered transformations
From legacy to cloud foundations, with architecture decisions proven by delivery.
- Since 2025CurrentFreelance
Confidential client
Real-estate asset management
Took over a poorly documented system and modernised it around an AI platform. Years of technical debt cleared.
- 2020 – 2024Freelance
Essilor
Optics
Test platform rebuilt, validation cut from several days to a few hours. CI/CD industrialised across about a hundred projects.
Validation that used to take several days now fits into a few hours, without ever putting production at risk.
Head of the test platform - 2018 – 2020via SoftFluent
Saretec
Insurance
Azure migration with replatforming and monitoring. Costs framed with FinOps.
- 2016 – 2018via SoftFluent
Carrefour
Retail
Global pricing platform (millions of prices), processing time brought down from minutes to seconds.
- 2010 – 2016
In System
Sage integrator
Custom development for its clients, from batch jobs to web portals.
- Dessange InternationalLuxury
Extranet and purchasing platform for franchisees. A custom remote-maintenance tool, deployed across 400 salons. A shared base with Camille Albane and the US brand Fantastic Sams.
- BPCEBanking
Data integrations and a tool for managing wealth portfolios.
- CNRS InnovationResearch
Management of the patent and licence portfolio from public research.
- StannahIndustry
Operations platform: technicians, logistics, parts and billing. Legacy VB6 migrated to .NET.
Expertise
AI & augmented engineering
Cloud & continuous delivery
Architecture & data
Identity & access (IAM)
Also comfortable with
Contact
Let's talk about what's stuck
Product scoping, architecture review, team reinforcement or development. Write to me with the context, even if it's still incomplete.
Emailsebastien.gissinger@gissit.comPhone+33 9 72 15 94 24LinkedInlinkedin.com/in/sgissinger
LocationAntony, near Paris · France