Hewlett Packard Enterprise

Supporting AI infrastructure pre-sales and business development across Central Europe for HPE Private Cloud AI (PCAI), an enterprise-ready AI platform co-developed with NVIDIA. I help customers define AI strategies, implement MLOps workflows, and scale model deployment on optimized hybrid architectures.

Key Contributions & Achievements:
→ Driving Pre-Sales Across Central Europe
Worked with business and technical stakeholders to address AI adoption challenges, demonstrate value, and support strategic deals.

→ Shaping AI Business Development
Helped define PCAI business strategy with partners and local teams, aligning with market needs and identifying new opportunities.

→ Strengthening Partnerships with NVIDIA and the AI Ecosystem
Led regional co-selling, customer engagements, and sales enablement with NVIDIA and channel partners.

→ Bridging the Gap Between AI Strategy and Execution
Guided organizations in creating AI roadmaps, enabling scalable infrastructure for model deployment and MLOps.

→ Educating and Enabling Teams
Delivered AI enablement sessions to internal sales and partner teams to strengthen regional go-to-market capabilities.

Solution Architect - HPE Private Cloud AI

Supported the go-to-market of HPE Ezmeral in Central Europe as a pre-sales engineer. Acted as a technical advisor to sales teams, partners, and customers, helping to demonstrate the platform’s value in modern data and AI environments. Contributed to field enablement, opportunity qualification, and feedback into product strategy.

Key Contributions & Achievements:
→ Sales Support & Technical Enablement
Worked with country architects and partners to explain Ezmeral’s capabilities, answer technical questions, and help close complex opportunities.

→ Hands-On Expertise in Data & AI
Developed practical knowledge in machine learning, data engineering, and Kubernetes to advise clients and support solution design.

→ Strategic Sales Collaboration
Aligned with account and sales teams to refine positioning, leverage competitive insights, and shape customer proposals.

→ Product & Market Feedback Loop
Collaborated with the product team to relay customer needs and influence the platform roadmap.

→ Regional Process Development
Helped define tools and best practices to support consistent pre-sales execution across the region.

Solution Architect - HPE Ezmeral - Data & AI SW Platform

Managed the design and delivery of large-scale infrastructure outsourcing and transformation projects for IBM’s major French clients. Led cross-functional solution teams, developed client proposals, and drove strategic account engagements with a focus on service innovation and profitability.

Key Contributions & Achievements:
→ Enterprise Solution Design
Led architecture for complex infrastructure services, including managed services and project-based IT transformations.

→ Client & Deal Ownership
Engaged directly with customers and senior stakeholders to understand business needs, tailor proposals, and support deal closure.

→ Service Definition & Contracting
Defined solution scopes, wrote statements of work, and managed customer expectations throughout project lifecycles.

→ Technology Enablement
Worked on deals involving hybrid cloud, big data, analytics, and emerging technologies like blockchain.

→ Innovation & IP Contribution
Submitted multiple patent ideas and contributed to IBM internal innovation initiatives.

→ High-Impact Engagements
Delivered over 50 major IT infrastructure deals across sectors, ensuring alignment with client business goals while driving margin for IBM.

Technical Solution Manager | Lead Solution Architect

Led the design and delivery of infrastructure solutions across Nordic and French accounts as a pre-sales architect. Translated complex technical needs into viable solutions using platforms like SAP, VMware, Unix, MS SQL, and Oracle. Acted as a trusted advisor to clients and led multi-disciplinary teams to deliver high-value proposals and implementations.

Key Contributions & Achievements:
→ End-to-End Solution Design
Architected technical solutions aligned with business goals, incorporating hybrid technologies to meet scalability, performance, and cost targets.

→ Client Engagement & Deal Support
Partnered with clients throughout the sales lifecycle, from discovery to proposal, to shape solutions and secure over €10M in business.

→ Leadership of Cross-Functional Teams
Managed internal and external expert teams, ensuring solution alignment, delivery feasibility, and clear documentation.

→ Profit-Driven Execution
Balanced client value with IBM’s profitability targets to craft win-win engagements that met margin expectations.

→ Complex Project Closure
Led the successful signing and planning of 100+ major IT service contracts, helping solidify IBM’s footprint in the region.

Infrastructure Services Solution Manager | Solution Architect

Alongside my main technical role, I served as an internal instructor within IBM, delivering structured training programs focused on AIX and Power Systems technologies. I enabled the growth of technical talent across IBM by teaching structured courses, mentoring certification candidates, and sharing advanced knowledge.

Key Contributions & Achievements:
→ Instructor for Core & Advanced AIX Courses
Delivered instructor-led courses across a broad technical spectrum including AIX basics, virtualization, advanced administration, PowerHA, and performance tuning.

→ Certification Mentorship
Guided colleagues through preparation for IBM Power Systems certifications, offering one-on-one coaching and structured learning paths.

→ Content Development & Customization
Adapted existing training materials to fit internal needs and real-world use cases, making complex topics accessible to all experience levels.

→ Cross-Team Enablement
Built strong relationships across departments by sharing expertise and creating a culture of continuous learning.

→ Recognition as a Trusted Knowledge Leader
Became a go-to expert for Power Systems training within the Brno office, reinforcing IBM’s internal capability development initiatives.

IBM internal Teacher

Directed large-scale technical infrastructure upgrades for French clients, managing project delivery across SAP, Oracle, Power Systems, and AIX migrations. Coordinated local and remote teams while maintaining operational continuity and ensuring alignment with customer expectations.

Key Contributions & Achievements:
→ Modernization & Migration Leadership
Oversaw Oracle and SAP upgrades, AIX migrations, and operational transitions from Solaris; improving reliability and performance.

→ Technical Oversight of Power Systems
Managed PowerHA clusters and Ethernet infrastructure, ensuring system uptime and optimized resource usage.

→ Crisis Management & Incident Response
Led a high-impact LDAP system recovery project with daily C-level reporting, restoring critical services under pressure.

→ Hands-On Infrastructure Buildouts
Designed and implemented server systems, ensuring optimal connectivity, redundancy, and scalability.

→ Collaborative Delivery Execution
Worked closely with diverse teams to coordinate technical delivery across multiple accounts and domains.

Technical Project Leader | UNIX Subject Matter Expert

In parallel with my role as Technical Project Lead, I assumed responsibility for coordinating the AIX release process for multiple French enterprise clients. Ensured smooth, compliant, and timely execution of software releases across a complex technical landscape.

Key Contributions & Achievements:
→ Release Calendar Ownership
Planned and managed the annual release schedule for AIX environments, aligning with client business needs and system maintenance windows.

→ End-to-End Release Coordination
Led release execution from pre-approval through verification and go-live, ensuring technical completeness and quality assurance at every stage.

→ Stakeholder Communication & Alignment
Acted as the bridge between delivery teams and clients, facilitating coordination and issue resolution throughout the release lifecycle.

→ Quality & Documentation Oversight
Maintained comprehensive documentation, testing evidence, and audit trails to support operational compliance and process improvement.

→ Dual-Role Management
Successfully balanced responsibilities across both technical delivery and process governance roles, showcasing versatility and reliability.

Release Process Coordinator

Atos

Acted as the lead Unix expert for a major client operations transition from IBM to Atos. Designed infrastructure, led technical migrations, and coordinated system standardization across Linux and AIX environments.

Key Contributions & Achievements:
→ Infrastructure Design & Delivery
Architected client-specific Unix environments, ensuring alignment with performance and security standards.

→ System Migration Execution
Led Linux-to-server transitions and AIX server deployments to meet business and technical transformation goals.

→ Automation & Scripting
Developed tools and scripts to streamline administrative tasks and increase operational efficiency.

→ Security & Compliance Hardening
Applied updates and configuration changes to align with internal and client-specific security requirements.

→ Stakeholder Communication
Provided regular progress updates to both client and internal stakeholders, ensuring visibility and trust throughout the engagement.

Subject Matter Expert

Progressed from Unix Administrator to Account Technical Leader, overseeing critical IT operations for major French accounts. Managed system performance, led knowledge transfers between international teams, and ensured reliable service delivery in complex environments.

Key Contributions & Achievements:
→ Operational Oversight of Unix Systems
Monitored, maintained, and optimized AIX and Unix environments to ensure stability and uptime for critical client systems.

→ Leadership in Knowledge Transfer
Led transitions of Unix operations from IBM France to IBM Poland, and later from Poland to IBM India, ensuring seamless continuity and capability retention.

→ Incident Management & Support
Handled high-severity technical incidents, often during off-hours, resolving issues under pressure with minimal disruption.

→ Client Relationship Support
Acted as a technical point of contact for key clients, aligning operations with business needs and building long-term trust.

→ Career Progression through Delivery Excellence
Promoted from admin role into leadership due to consistent delivery, problem-solving skills, and team collaboration.

Account Technical Leader

Started as a system operator and quickly progressed into a leadership role, overseeing a team responsible for daily monitoring and support of IT systems. Played a key role in transitioning operations to a new IBM delivery center in Poland and leading the new team to success.

Key Contributions & Achievements:
→ Operational Excellence
Monitored batch jobs, reports, and system health checks to ensure daily tasks ran smoothly with minimal failures.

→ Team Leadership & Coordination
Led a team of 16 operators, overseeing shift planning, quality control, and internal communication.

→ Transition Project Execution
Managed the operational transition from the Czech Republic to IBM Poland, relocating for three months to train the new team and ensure operational readiness.

→ Mentorship & Skills Development
Supported the growth of junior team members, helping them ramp up quickly and contribute effectively to the new delivery center.

→ Foundation of a Technical Career
Laid the groundwork for a successful career in enterprise IT through dedication, adaptability, and continuous learning.

Computer Operator | Team Leader

Hi! I'm Vincent, fifteen years ago, I moved to Brno (Czechia) for ‘just a year.’ Now I spend my days at HPE trying to turn AI buzzwords into business value without putting anyone to sleep.
At home, I’m usually explaining to my kids that AI won’t steal their toys. At work, I help companies build Private Cloud AI solutions they can actually trust.
Over the past 15 years, I’ve helped large companies to improve their business efficiency through successful digital transformation. 
Today, I lead technical strategy and solution architecture for key enterprise accounts. In addition to designing AI and data-driven solutions, I actively contribute to business development, opportunity qualification, and go-to-market planning. I work closely with sales teams, partners, and customers to shape solutions that address real business needs and position them for long-term success.

Multi-year experience focusing on the following areas:
- Solution Architecture, 
- Customer/Account Management, 
- Program/Project Management, 
- Service Delivery & Incident Management, 
- Leadership, 
- Global Deal Execution with Multi-Cultural Teams, 
- Organizational Change Management 
- Maintenance of business continuity, 

Technology Exposure:
- AI/ML: MLOps (MLflow, Kubeflow, Ray), LLMs, Databricks (Certified ML Engineer) and deployment on-premises and hybrid infrastructures;
- Data Engineering: SQL, Spark, Airflow, Hadoop, MapR/HPE Ezmeral Data Fabric Software, AWS (Redshift, S3, EMR);
- Cloud Native: Kubernetes (CKAD), ArgoCD, Git, Docker; 
- Cloud: IBM, Amazon Web Services & Microsoft Azure (Microsoft Certified);
- Physical Server & Storage Configuration (x86, p-Series, Mainframe, etc.) and HPE (Compute, Storage, Hyperconverged)

If you're looking for a professional to lead your next data-driven project, explore innovative AI/ML solutions, or drive your organization's digital transformation, feel free to connect with me. I look forward to collaborating on exciting ventures that push the boundaries of technology and business.


[Keywords: Data & AI, Solution Architect, Digital Transformation, Account Management, Program/Project Management, Leadership, Global Deal Execution, Cloud Native, AI/ML, AWS, Azure, Kubernetes, Data Engineering]

Vincent Charbonnier

Passionné par la technologie en général, mon profil de Data Full Stack me permet d'intervenir sur l'ensemble de la chaîne de valeur d'un projet, que ce soit au niveau technique ou business, en sensibilisant le métier et les collaborateurs aux enjeux de la data.

À plusieurs reprises, j'ai contribué à lancer des projets à partir de zéro, ce qui m'a permis d'acquérir des compétences et une compréhension solide de l'écosystème data en entreprise.

En ce qui concerne le Data Engineering, j'ai eu l'occasion de mettre en place des infrastructures ELT avec les technologies de la Modern Data Stack (Airbyte, dbt, Redshift, Dagster, ...).

Du côté du Machine Learning, j'ai déployé des solutions de gestion du cycle de vie des modèles ainsi que leur mise en production.
Récemment, j'ai assumé le rôle de lead Data sur un projet de mise en production de LLM avec une architecture entièrement serverless.

Je suis spécialisé dans AWS et certifié Solution Architect, ce qui me permet dans chacune de mes expériences de développer des infrastructures sécurisées, robustes et scalable en fonction des besoins.

Adeptes des bonnes pratiques, je mets en place dès que possible des tests unitaires et d'infrastructure, ainsi que de l'Infrastructure as Code avec des frameworks comme Terraform ou Pulumi. Le tout est le plus souvent déployé via du CI/CD sur GitLab ou GitHub.

Ma passion pour la technologie me pousse à faire constamment de la veille, mais aussi à rédiger des articles que je partage avec le plus grand nombre sur mon compte Medium

Jérémy Surget

La plupart de ceux qui disent "Nous entraînons notre IA" mentent


C'est généralement du context engineering ou du prompt bien ficelé, pas de l'entraînement.

Le moyen le plus efficace d'"entrainer" réellement un LLM à moindre coûts c'est de faire du LoRA (Low Rank Adaptation)

Le principe est simple :
- On ne touche pas aux milliards de paramètres du modèle.
- On injecte deux petites matrices (A et B) qui viennent "modifier" le comportement du modèle sur les couches clés.

Le modèle de base reste gelé et on a juste à entrainer ces matrices 

En pratique ça donne :
→ 0.1% des paramètres à entraîner vs 100%
→ Quelques Mo de poids au lieu de centaines de Go
→ Besoin d'un seul GPU au lieu d'un cluster
→ Zéro surcoût à l'inférence (les matrices fusionnent avec le modèle)

Le LoRA est par exemple beaucoup utilisé sur Hugging Face avec des milliers de modèles fine-tunés par la communauté.

Sans LoRA, fine-tuner un LLaMA 7B est réservé aux boîtes avec d'énorme budget
Avec LoRA n'importe qui peut le faire sur une RTX 3090

(bon en vrai, ca m'arrive aussi de prendre le raccourci de dire que l'IA est entrainée)

Co-founder DiscutAI & Freelance AI Engineer

lora

1 million.


On vient de dépasser 1 million de conversations sur DiscutAI​​

Il y a 10 mois, on codait les soirs et les week-ends avec Paul Olivier Ahoua​ et Bertin Jean-François TETCHI

Aujourd'hui, je me lance à 100% sur le projet.

𝗗𝗶𝘀𝗰𝘂𝘁𝗔𝗜, 𝗰'𝗲𝘀𝘁 𝗾𝘂𝗼𝗶 ?

Une plateforme d'agents IA pour le service client sur WhatsApp.

L'objectif : répondre à 100% des messages clients en moins de 10 secondes.
Jour et nuit.
Sans embaucher.

Mais pas juste répondre. Agir.
→ Prendre une commande
→ Générer un lien de paiement
→ Créer un ticket intelligent
→ Se connecter au CRM, à l'ERP, à n'importe quelle API

𝗢𝘂̀ 𝗼𝗻 𝗲𝗻 𝗲𝘀𝘁 𝗮𝘂𝗷𝗼𝘂𝗿𝗱'𝗵𝘂𝗶 :

✅ 6 500€ de MRR
✅ 2000 messages traités par jour
✅ 38 clients (PME, startups, e-commerce, et quelques boîtes de +1000 employés)

𝗖𝗲 𝗾𝘂'𝗼𝗻 𝘃𝗶𝘀𝗲 :

100 clients le mois prochain.

Et continuer à construire le service client là où sont vraiment les utilisateurs.
Sur WhatsApp.

Des agents IA qui font vraiment le job.
Pas juste des chatbots qui récitent une FAQ.

On ne fait que commencer 🚀

Voici les principes qu'on applique systématiquement quand on implémente un tool pour nos agents IA👇

Chez DiscutAI​, nos agents sont en première ligne : ils parlent directement aux clients de nos utilisateurs. C'est pour ca que lorsqu'on implémente un tool on a pas le droit à l'erreur

→ 𝗡𝗲 𝗷𝗮𝗺𝗮𝗶𝘀 𝗳𝗮𝗶𝗿𝗲 𝗰𝗼𝗻𝗳𝗶𝗮𝗻𝗰𝗲 𝗮𝘂 𝗟𝗟𝗠 𝗽𝗼𝘂𝗿 𝗹𝗲𝘀 𝗽𝗮𝗿𝗮𝗺𝗲̀𝘁𝗿𝗲𝘀
Tout ce qui entre dans un tool passe par une validation stricte côté code. Pydantic est notre gardien. Un LLM peut mal formater une date, notre code ne doit pas laisser passer ce genre d'erreur

→ 𝗜𝗻𝗷𝗲𝗰𝘁𝗲𝗿 𝘂𝗻 𝗺𝗮𝘅𝗶𝗺𝘂𝗺 𝗱'𝗶𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗱𝗮𝗻𝘀 𝗹𝗲 𝘁𝗼𝗼𝗹, 𝘀𝗮𝗻𝘀 𝗽𝗮𝘀𝘀𝗲𝗿 𝗽𝗮𝗿 𝗹𝗲 𝗟𝗟𝗠
Si un paramètre est connu (date du jour, nom du user), on ne demande pas au LLM de le passer en paramètre, on l'injecte directement. Ca évite les soucis et on a moins d'erreur sur l'appel des tools 

→ 𝗠𝗲𝘀𝘀𝗮𝗴𝗲𝘀 𝗱'𝗲𝗿𝗿𝗲𝘂𝗿 𝗲𝘁 𝗱𝗲 𝘀𝘂𝗰𝗰𝗲̀𝘀 𝗰𝗹𝗮𝗶𝗿𝘀
Si un tool fail à cause d'un mauvais paramètre, le LLM doit pouvoir se corriger. Si le tool réussi, le LLM doit savoir quoi dire au user. Dans tous les cas, plus le retour du tool est clair, meilleur sera le LLM

→ 𝗝𝗮𝗺𝗮𝗶𝘀 𝗱𝗲 𝗱𝗼𝗻𝗻𝗲́𝗲𝘀 𝘀𝗲𝗻𝘀𝗶𝗯𝗹𝗲𝘀 𝗲𝗻 𝗿𝗲𝘁𝗼𝘂𝗿 𝗱𝗲 𝘁𝗼𝗼𝗹
Un tool qui renvoie un email ou un numéro de téléphone ? C'est une porte ouverte. Si quelqu'un se fait passer pour un utilisateur, l'agent pourrait lui servir ces infos sur un plateau. Least privilege doit être appliqué à l'extrême.
Nos tools n'ont accès qu'à ce dont ils ont strictement besoin. Lecture seule, sur un sous-ensemble restreint de la base. Jamais d'accès global "au cas où" et surtout, on ne retourne jamais d'informations sensibles pour éviter que le LLM ne puisse les utiliser

→ 𝗗𝗲𝘀 𝗹𝗼𝗴𝘀 𝗰𝗹𝗮𝗶𝗿𝘀 𝘀𝘂𝗿 𝗰𝗵𝗮𝗾𝘂𝗲 𝗲𝘅𝗲́𝗰𝘂𝘁𝗶𝗼𝗻.
Quand quelque chose plante (et ça arrive), on veut savoir exactement quel tool a été appelé, avec quels paramètres, et ce qu'il a renvoyé.


Les LLMs sont impressionnants, mais ils restent imprévisibles. Notre job c'est de construire le système autour pour que leurs erreurs ne deviennent jamais les erreurs de nos clients.

On développe une architecture pensée pour limiter les dégâts quand le modèle se trompe, parce qu'il se trompera

taplio

The Battle

Vincent Charbonnier

Jérémy Surget

Followers

Posts

Engagements

Likes

Performance ScoreJérémy Surget's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Vincent Charbonnier

Jérémy Surget

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Jérémy Surget's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreJérémy Surget's LinkedIn Performance