Thirty years of technology leadership in Tunisia, Qatar, the United Arab Emirates, Germany and France. I work in four ways: in technology leadership roles, as a consultant, as a trainer, and in my lab applying AI to quantitative trading.
Mohamed Ben Lakhoua · Infrastructure & Cloud Architect · 30 years of critical platforms
NATO · European Patent Office · Stuttgart Stock Exchange · Abu Dhabi e-Government · a major French banking group
For a CTO, VP Engineering or Platform, or Director Infrastructure & Cloud role: thirty years building, leading and running critical platforms, with teams spread across countries and a P&L to deliver.
The Sandbox · Paris
Global consumer platform, 8M+ users; reporting to the CTO, with Executive Committee exposure.
]init[ AG · Berlin
Public sector and regulated financial services: NATO, European Patent Office, Stuttgart Stock Exchange.
]init[ AG · Abu Dhabi
e-Government programme: a sovereign Tier-4 platform connecting 50+ ministries, agencies and banks, run 24/7.
Qatar Foundation · Doha
National digital services under formal service commitments.
Letters of recommendation and thanks from ministers and public institutions in Europe and the Middle East, available on request.
Engagements that start from the strategic context, translate it into architecture, then into measurable service commitments. On site or hybrid in the Paris region, or remote.
Consulting
A critical platform has to grow, stay compliant and keep its commitments while regulation, AI and costs all change at once. These requirements often conflict: architecture is where they are arbitrated, and measurement is what makes them verifiable.
DORA has applied since January 2025, Cyber Resilience Act reporting since September 2026, NIS2 is arriving in France with the loi Résilience, and the AI Act is coming into application in stages. Regulators expect controlled risk and suppliers under control.
A few clouds and a few model providers concentrate the risk: you need to know what holds when one of them fails, and be able to exit.
AI agents now act on systems: they require guardrails, evidence and suitable service levels.
GPUs, inference and tokens add to the cloud bill: without unit cost and trade-offs, they drift.
Where the data lives, who can access it, and under which law.
Service processes built on shared objectives, and metrics that show whether you are improving.
Consulting
Each area can be engaged on its own; they reinforce each other when combined.
Designing the target platform and scaling it: public cloud, private cloud and hybrid.
Holding up when a link fails, and proving it to auditors and regulators.
Operations built on measured objectives rather than on the urgency of the moment.
AI agents that collect, calculate, detect and draft; decisions stay human.
Consulting
Cloud migration, compliance or service quality: the approach is the same. Start from what the platform actually delivers, then commit. I work at every stage, from scoping to governance.
Assessment of the current state, service and dependency mapping, business and regulatory requirements, DORA and ISO gap analysis.
Target architecture, resilience and redundancy, capacity tiers, service level model.
Architecture documents, BCP and DRP, SLA, OLA and UC sheets, RACI and compliance matrices.
Monitoring and metrics, infrastructure as code, dashboards, AI automation.
Service and architecture reviews, audits, continuous improvement, team upskilling.
Consulting
A concrete example of the method. An SLA is only enforceable when it rests on achievable OLAs and UCs, themselves backed by reliable measurement. Pick an availability: the whole chain recalculates.
Illustrative assumptions: SLO = SLA with a 50% margin on the error budget; 3 internal teams and 1 supplier in series; 30-day month.
Commitments are built from the bottom up; penalties flow from the top down.
What is promised and billed: availability, restoration times, credits and penalties.
The internal objective, stricter than the SLA, and its error budget.
Run, platform, network, security: what each team commits to deliver for the service.
Cloud, telecom, software vendors, data centre: the underpinning contracts that must cover the SLO.
Indicators drawn from existing monitoring (Zabbix, Dynatrace, Centreon, Prometheus), with an auditable calculation method.
Observed history: actual availability, incidents, resilience tests.
The examples recalculate with the availability selected in the calculator. 99.9%
B = (1 − SLO) × periodThe downtime allowed over the period (here 30 days = 43,200 min).
ExampleWith an SLO of 99.95%: (1 − 99.95%) × 43,200 min = 22 min per month.
SLO = 1 − (1 − SLA) × (1 − m)A margin m = 50% of the SLA budget is kept to react before any penalty.
ExampleSLA 99.9%: SLA budget 43 min per month, half of which is kept ⇒ SLO 99.95%.
A = A₁ × A₂ × … × Aₙ ⇒ Aᵢ ≥ SLO^(1/n)If 3 teams and 1 supplier are all required (n = 4), each must do better than the SLO.
Example99.95%^(1/4) = 99.988% for each of the 4 links (3 OLAs + 1 UC).
A = 1 − (1 − a)ᵏTwo independent instances at 99.5% each give 99.9975%: architecture is what lets you promise more.
Example2 × 99.5% in parallel ⇒ 1 − 0.005² = 99.9975%.
SLI = good events / valid eventsMeasured over a rolling 30 days. Example: 999,600 successful requests out of 1,000,000 = 99.96%.
Example999,600 / 1,000,000 = 99.96%.
Consulting
Monitoring, capacity, cost, compliance, service levels: once the rules are set, the chain can run on its own. AI agents collect, calculate, detect and draft; decisions and commitments stay human.
Metrics, incidents, changes, cloud inventory and costs, extracted automatically from monitoring, ITSM and cloud consoles.
Availability, error budgets, capacity, costs and DORA DevOps metrics, recalculated continuously by tested and auditable code.
Capacity or cost drift, a budget burning too fast, an OLA or UC putting the SLA at risk, a compliance gap, a missing measurement.
Drafts of architecture documents, SLA, OLA and UC sheets, post-mortems, compliance registers and evidence.
Dashboards, the monthly service review and the compliance report generated automatically, ready for sign-off.
The topics IT teams need to master this year: AI moving into production, regulation coming into application, costs to control. In-house, in French, English or Arabic, built on your own services.
Metafive 360° is my lab: an algorithmic trading platform where several AI agents design, test, review and deploy code under written governance. Trading is the proving ground; the object of the research is the governance of multiple AI agents on a system where every mistake is measured.
What this lab brings to my engagements: hands-on practice of agentic AI in production, agent governance and statistical rigour, reused directly in consulting and training. The agent-constitution governance framework (open source)
Personal research and engineering project. This site provides no investment advice, no signals and no third-party asset management. Trading leveraged products carries a high risk of capital loss.
Thirty years of production platforms where downtime is expensive.
I am Mohamed Ben Lakhoua, Infrastructure & Cloud Architect and founder of Metafive.ai. Since 1995 I have designed and governed critical platforms in Tunisia, Qatar, the United Arab Emirates, Germany and France, and led organisations of 25 to 144 people across four continents.
Today I support infrastructure leaders on the architecture of their platforms, their resilience, their compliance and the quality of their operations, and I train teams to sustain them. SUP'COM engineer; I work in French, English and Arabic.
A first 30-minute conversation to understand your context: architecture, compliance stakes, commitments, deadlines.
Based in Paris – La Défense. On-site, hybrid or remote engagements.