Back to the atelierThe team

About the author

In depth

Louay Belkhamsa

Engineer and tech advisory consultant, writing on AI strategy and change.
Cloud and GenAI engineer, advising on AI strategy.

Louay is an engineer trained at Télécom SudParis, working as a tech advisory consultant on generative AI. Inside a consultancy he has helped build a sovereign GenAI platform, guided client teams through AI adoption, and modelled cloud migration scenarios for large French companies.

Spiresen is where he takes the questions that work raises and pushes them further: AI strategy and change, argued from independent sources and tested in projects.

Louay builds GenAI systems on AWS and Azure: agentic Python backends, cloud architecture and FinOps. He has taken internal GenAI projects from experiment to production, and works daily with the CEO, CTO and AI lead on how the company adopts AI and changes with it.

Spiresen is where that practice meets strategy: AI-oriented approaches, tested in projects whose cost, quality and time are tracked.

Work so far

  • Sep 2025 – Aug 2026

    Tech advisory consultant, apprenticeship

    Timspirit · IT consultancy · missions at large French companies (CAC 40)

    Apprenticeship alongside the CEO, the CTO and the AI lead, helping the group get to grips with AI and adopt it.

    Building the company’s AI offer

    Helped build the consultancy’s own sovereign AI platform, working closely with business teams to choose and deliver new AI use cases. Framed proofs of concept with clear success measures, and used the results to guide product decisions. Worked with HR and a business manager on specific cases so the tool answered their real needs.

    AI adoption and change

    Ran interviews and workshops with business teams, assessed GenAI tools and advised on adopting them, and turned AI topics into something non-technical colleagues could act on. Kept an AI watch for the tooling committee (CEO, CTO, AI lead) and wrote a monthly AI newsletter for the group. Advised the sourcing consultants of another business unit on using AI to assess client tenders, running use-case sessions with consultants and sales.

    Client missions

    For a cosmetics group: compared hosting scenarios and cloud suppliers, modelled the financial case for each migration option, and gathered rough cost estimates directly from providers to support contract renegotiation. For an insurer: defined cloud architecture standards that reconcile group governance with the constraints of its subsidiaries.

    Employer
    Timspirit, an IT consultancy, with missions at large French companies (CAC 40 level).
    Platform
    Distributed, scalable AWS infrastructure (CDK, ECS, DynamoDB, S3) for a sovereign chatbot on an LLM/SLM stack.
    Backend
    Python (FastAPI) with agentic workflow orchestration (A2A).
    Multi-cloud
    Azure AI (Foundry, AI Search) for LLM inference and vector search. RAG, document generation, agent-to-agent, SharePoint integration.
    Quality
    Clean Code standards (DDD, TDD, PEP 8, Ruff) and a human-in-the-loop agentic dev workflow. 700+ tests, 14 production releases, Agile delivery.
    Security, ops
    Zero-trust (IAM, STS, KMS), multi-environment CI/CD, CloudWatch logs, metrics and alarms.
    Advisory
    Cloud hosting scenarios with 5 providers compared on service equivalence and cost. Cloud architecture standards on the AWS Well-Architected Framework, and FinOps optimisation processes. Needs gathering with business teams, an AI watch and a monthly newsletter.
  • Jul – Aug 2025

    Research intern, SAMOVAR Lab

    Modelled the energy use of cloud applications, to support sustainability and cost optimisation.

    Model
    Energy consumption of cloud applications across compute, memory and network.
    Method
    Containerised environments (Linux, Docker) to simulate and profile distributed workloads.
    Aim
    Sustainability, and cost-versus-performance optimisation.
  • Jul – Sep 2023

    QA data analyst intern, Vneuron

    Defined measures for software quality and replaced a manual reporting task with automated reports, saving one to two days per delivery.

    Metrics
    KPIs to structure and evaluate the QA procedure in an Agile cycle.
    Automation
    Automated Jira dashboards replacing a manual task: 1 to 2 days saved per delivery.

Stack

Generative & agentic AI
LLM/SLM · RAG · Agent-to-Agent (A2A) · Azure AI Foundry and AI Search · AWS Bedrock · prompt and context engineering
Dev & quality
Python · FastAPI · DDD · TDD · Clean Code (PEP 8, Ruff) · automated tests
Cloud & DevOps
AWS (CDK, ECS, S3, DynamoDB) · Azure AI · Terraform · Kubernetes/EKS · CI/CD · Docker · IAM and zero-trust · CloudWatch

Projects

  • 2026 –

    On AI short-termism

    A four-part article series on why companies adopt AI faster than they can turn it into value, and on bounded rationality in the age of AI.

    Format
    A series of four articles on why companies adopt AI faster than they can turn it into value.
    Idea
    Herbert Simon’s bounded rationality, applied to AI adoption, with concrete figures.
  • 2026 –

    Spiresen Saga

    A personal project on planning AI-assisted development so it costs less before the first line of code runs.

    Goal
    Turn scattered research into FinOps standards for AI-assisted development, to cut spend and risk.
    Output
    Sourced decision records, 72 engineering rules, an agentic topology and strategies for coding, deployment and maintenance.
    KPI
    TCO per maintained feature.

Certifications

  • AnthropicClaude Certified Architect, Foundations
  • AWSAI Business Strategist, Early Adopter
  • AWSCloud Practitioner
  • CambridgeC2 English (CAE)

Languages

  • Englishnative
  • Arabicnative
  • Frenchbilingual

Education

  • 2024 – 2026

    Télécom SudParis

    Engineering diploma in IT, specialised in the integration and deployment of information systems.

    Degree
    Engineering diploma in IT, specialised in integration and deployment of information systems.
  • 2022 – 2024

    SUP’COM, Tunis

    Engineering in information technologies.

    Degree
    Engineering in information technologies. Rank 6 of 147.
  • 2020 – 2022

    IPEST, Tunisia

    Preparatory classes (PCSI-PC).

    Programme
    Preparatory classes PCSI-PC. National rank 76 of 1,400.
  • 2016 – 2020

    Lycée Ibn Abi Dhiaf

    Mathematics baccalaureate with high distinction.

    Result
    Mathematics baccalaureate, 18.10 / 20. National rank 177 of 9,677 (top 2%).

Also

  • Project lead, Junior Enterprise
  • First prize, crowdfunding hackathon, Tunisia
  • First place, pitch competition, Leading Sup’Com