Hugues Martin
Cagnes-sur-Mer, Provence-Alpes-Côte d’Azur, France
583 abonnés
+ de 500 relations
À propos
I am passionate about actively listening and understanding customer Application Security…
Activité
583 abonnés
-
Hugues Martin a partagé ceciA great post about secret leakage in the SDLC and preventing it efficiently without slowing developers down.Hugues Martin a partagé ceciSecret leaks are doubling every year. How do we ensure our capacity to protect software grows with our capacity to create it? In collaboration with Microsoft Applied Sciences, we built a model that has the potential to more than double the secrets GitHub stops before they leak. ✨ Today, were announcing our purpose-built model for leaked secrets detection. The fine-tuned ModernBERT classifier detects unstructured secrets from surrounding code context, while balancing precision, latency, throughput, and cost. It's more precise than our LLM-based pipelines, while also being very fast! Current data shows the model evaluating a whole set of candidates at push time in under 2 ms. 👏 We're bringing it to push protection, in addition to other native developer surfaces mentioned in our blog post, linked below. Thank you Batu Ozturkler, Shengyu Fu, Simona (Zhehui) Liao and the Applied Sciences team for the research partnership. Robert Bolender, Krystal Mitchell, and Ashwin Mohan for technical leadership, and Jarrod Schnapper for our experimentation pipeline. John Gregson for business planning. Colten Woo for marketing. Michael Badgett, Daniel Dunn, Jessica Schumaker, Adriana Babakanian, Gleb Irovich and Aakash Shah for bringing our model into developer surfaces. Prevention scales with compute, but remediation still scales with people. We owe the developer community the same progress in protecting software that we are delivering in producing it. https://lnkd.in/gVV7i3YV
-
Hugues Martin a partagé ceciI'm happy to share that I've just received the CISSP certification.Certified Information Systems Security Professional (CISSP) was issued by ISC2 to Hugues Antonio Martin.Certified Information Systems Security Professional (CISSP) was issued by ISC2 to Hugues Antonio Martin.
-
Hugues Martin a republié ceciHugues Martin a republié ceciVoici l'avant dernière newsletter de AppScan avant les vacances, cette semaine Hugues Martin nous présente la sécurisation des agents IA via IAST ! N'hésitez pas à nous contacter pour plus de détails ! si vous souhaitez que nous traitions un sujet dans les prochaines Newsletter, n'hésitez pas ;)
-
Hugues Martin a republié ceciHugues Martin a republié ceciCredi +1 ;) Cette semaine Hugues Martin nous propose de revoir ou de voir les derniers webinaires en français réalisés par l'équipe AppScan France ! si vous avez un peu de temps, dans les transports ou pendant les premiers matchs de la CM ;), N'hésitez pas ! et si vous avez des questions, nous sommes à votre service !
-
Hugues Martin a republié ceciHugues Martin a republié ceciC'est ... Jeudi ! hé oui, la newsletter AppScan a un peu de retard cette semaine ! Cette semaine, Hugues Martin précise les dernières innovations dans notre moteur SAST et pourquoi cette technologie est toujours une brique de base d'un programme AppSec ;)
-
Hugues Martin a republié ceciHugues Martin a republié ceciJ-2 jours avant le CRiP Symposium ! 📅 En tant que décideurs IT, vous êtes constamment à la recherche du juste équilibre entre accélération business et maîtrise des risques. C'est le cœur de mission de HCL Software, et la raison pour laquelle nous serons présents à ce grand rassemblement de l'écosystème. Venir à notre rencontre au CRiP Symposium, c'est découvrir comment nous répondons concrètement à vos priorités opérationnelles : 🔹 Orchestrer avec agilité : Pilotez vos Workloads, Workflows métiers et Agent IA en temps réel et sur tous vos environnements grâce à notre plateforme cloud-native. 🔹 Sécuriser vos parcs : Réduisez votre surface d'attaque grâce à la remédiation automatisée et continue de vos endpoints. 🔹 Protéger vos applications : Bloquez les cyberattaques avant la mise en production avec les tests de sécurité applicative intégrés à votre chaine de développement. Rencontrons-nous ! Que ce soit autour d'un café ou lors d'une session d'échange, je serais ravi de partager nos retours d'expérience et d'évoquer vos défis du moment. 📩 Disponibilités en MP – n'hésitez pas à me contacter dès maintenant pour fixer un moment ! #CRiPSymposium #HCLSoftware #InnovationIT #Cloud #Cybersecurity #EndpointManagement #Networking
-
Hugues Martin a republié ceciHugues Martin a republié ceciFinding vulnerabilities has never been the hard part; fixing them is. HCL AppScan 360° v2.1 now introduces Agentic AI to close the gap. This leading-edge on-prem AppSec platform now integrates with RapidFix for: ▪️ Smart triage recommendations. ▪️ AI-generated code fixes. ▪️ Seamless integration into your Git workflow. Shrink your remediation timelines and keep your developers focused on building, not debugging security reports. Learn more: https://hclsw.co/xwthih #HCLAppScan #ApplicationSecurity #HCLSoftware #AppSec #DevSecOps #GenAI
-
Hugues Martin a republié ceciHugues Martin a republié ceciC'est Mercredi, le jour des enfants et ... de la newletter AppScan. Cette semaine Hugues Martin détaille les différents niveaux d'IA intégrés dans et avec AppScan pour plus d'efficacité et réduire les coûts sans réduire la qualité !
-
Hugues Martin a aimé ceciJanina Vinklere and Chris are spot on with the 10-80-10 rule. The best AI users get it. But too many underestimate, or simply skip, those critical 10%s. * First 10%: Direction. Like launching a rocket, a tiny error in trajectory and you miss the target entirely. This is where the real thinking happens. Get the objective, context or assumptions wrong, and even the best AI wont save you. * Last 10%: Ownership. AI is incredibly good at making things look right. And that's precisely the danger. It builds on what you give it, not necessarily what you forgot or got wrong. The result can look brilliant, reinforce your assumptions, and give you the illusion that you've done everything right. Yet behind that polished output may be major errors, omissions or complete nonsense. AI can do 80% of the work. But the remaining 20% determines whether that work is worth anything at all, or not. And that's where human judgment makes all the difference.Hugues Martin a aimé ceciMore or less exactly how I work now, just Chris Donnelly put it writing. What would you add? Or how are you using AI? I made this screenshot from his newsletter, and here is what screenshot couldn’t fit: “You check it, tighten it, and make sure what reaches the customer is high quality. I use these steps for almost everything I do now, so when someone new joins the team, they don't need to book time with me or chase five different people for answers. They just query Notion and get the context they need instantly. Simple doesn't mean easy, though. The hard part is making that 80% good enough to trust, and this is where people often fall short.”
-
Hugues Martin a aimé ceciHugues Martin a aimé ceciSecret leaks are doubling every year. How do we ensure our capacity to protect software grows with our capacity to create it? In collaboration with Microsoft Applied Sciences, we built a model that has the potential to more than double the secrets GitHub stops before they leak. ✨ Today, were announcing our purpose-built model for leaked secrets detection. The fine-tuned ModernBERT classifier detects unstructured secrets from surrounding code context, while balancing precision, latency, throughput, and cost. It's more precise than our LLM-based pipelines, while also being very fast! Current data shows the model evaluating a whole set of candidates at push time in under 2 ms. 👏 We're bringing it to push protection, in addition to other native developer surfaces mentioned in our blog post, linked below. Thank you Batu Ozturkler, Shengyu Fu, Simona (Zhehui) Liao and the Applied Sciences team for the research partnership. Robert Bolender, Krystal Mitchell, and Ashwin Mohan for technical leadership, and Jarrod Schnapper for our experimentation pipeline. John Gregson for business planning. Colten Woo for marketing. Michael Badgett, Daniel Dunn, Jessica Schumaker, Adriana Babakanian, Gleb Irovich and Aakash Shah for bringing our model into developer surfaces. Prevention scales with compute, but remediation still scales with people. We owe the developer community the same progress in protecting software that we are delivering in producing it. https://lnkd.in/gVV7i3YV
-
Hugues Martin a aimé ceciHugues Martin a aimé ceciYesterday I spoke at INTERPOL's 5th New Technologies Forum in Vienna, co-hosted with the Complexity Science Hub and the Bayerisches Staatsministerium der Justiz. About 100 people in the room, investigators and officials, from 45 INTERPOL member countries. A six-month-old French company on that stage. There are not many precedents for it, and I do not take the invitation lightly. My session was on Reddit as criminal infrastructure: where recruitment actually happens, what survives a deletion, and how a network that disappeared a year ago can still be mapped. The part I will remember is what happened afterwards. Investigators came up one after another with a case already in hand, asking whether the archive would have helped on it. Several of those conversations ran longer than the talk did. Thank you to 🔍️Vincent Danjean and the INTERPOL team for the invitation, and to everyone who stopped to question and challenge. I am sorry to have missed the second day. Back at it, with a longer list than I arrived with.
-
Hugues Martin a aimé ceciHugues Martin a aimé ceciJust earned my GitHub Foundations and GitHub Advanced Security certifications. Gotta know the basics! https://lnkd.in/eZfkTv6UGitHub Learn: Your Personalized Learning ExperienceGitHub Learn: Your Personalized Learning Experience
Expérience
Formation
Licences et certifications
Recommandations reçues
-
Membre LinkedIn
21 personnes ont recommandé Hugues
Inscrivez-vous pour y accéderDécouvrir plus de posts
-
Pierre Lalanne
Happy to see how far GitGuardian agents can go with tools such as Linkup! 🤝 For any company in the world, we can build a comprehensive picture of their publicly leaked secrets exposure, backed by in-depth investigation. 🔎 Congrats Alexis Begnez Alexandre Pradeilles Valentin de La Perraudière Samuel Guillaume 👏 And it's only the beginning!
-
ZeroToOne.AI
Our Principal Platform Engineer on the optimizations that keep our data infrastructure agile at scale. When you're processing behavioral signals across dozens of concurrent Spark pipelines, the platform either gets out of the way or it becomes the bottleneck. Manoj Babu and the team have made sure it's the former. This is what engineering excellence looks like from the inside. Blog in the comments.
1 commentaire -
Henri Tilloy
Root joining Aikido Security is acquisition number 5. And that product integration is French haute couture. Aikido spots CVEs, Root patches and distributes them autonomously. And with that, customers can safely rely on hardened images and libs, created on the fly. Security vendors used to brag about their detection capabilities and signal-to-noise ratio. But the future lies in automated fixes. Similar to how soft eng is moving from prompt to loop engineering. Security is shifting from detection+remediation to self-securing loops. This will soon be the new norm. https://lnkd.in/eB5G67C6
2 commentaires -
ELASTIC Project
New #ELASTIC Demo Series Episode: Propeller – Lightweight Orchestration Beyond Kubernetes We’ve released a new #demo showcasing #Propeller, a WebAssembly-based orchestrator designed to run application logic from the cloud all the way down to small edge and #IoT devices. Propeller uses SuperMQ for message distribution and deploys Wasm modules only when needed. This keeps the footprint small and makes it easier to run logic on devices with limited resources. In the demo, you can see how Propeller: 🔹 deploys Wasm functions across different environments 🔹 reacts to events through SuperMQ 🔹 evaluates conditions through a simple rules engine 🔹 keeps execution predictable on resource-constrained hardware Propeller is part of ELASTIC’s work on efficient and distributed execution for future 6G-ready systems. 🔗 Read full article: https://lnkd.in/dHJVKVZv 🎥 Demo video: https://lnkd.in/dCXrZhFA #ELASTICProject #6G #WebAssembly #EdgeComputing #Orchestration #OpenSource #IoT #Wasm Smart Networks and Services Joint Undertaking (SNS JU) Technical University of Crete Ericsson Telefónica Innovación Digital THALES SIX GTS FRANCE SAS Thales Cyber and Digital imec Ultraviolet Group Aalto University Lund University Abstract Machines Zentrix Lab Politecnico di Torino
-
Build5Nines LLC
Terraform didn’t just automate infrastructure — it changed how teams build it. 🚀 From multi-cloud provisioning to version-controlled, declarative workflows, HashiCorp Terraform became a core part of modern DevOps. Key takeaways: • Introduced in 2014 to unify infrastructure provisioning across providers • Popularized IaC with HCL, modules, state management, and plan/apply workflows • Helped teams manage AWS, Azure, GCP, Kubernetes, and on-prem from one codebase • Reached 1.0 in 2021, then faced licensing controversy in 2023 • OpenTofu emerged as an open-source alternative Read the full story to understand Terraform’s evolution, impact, and current place in the IaC ecosystem. 👉 https://lnkd.in/ehSjBDv3 #Terraform #IaC #DevOps #HashiCorp #CloudComputing
-
Teleglobal International
High AWS bills hurting your cloud ROI? Over the last few weeks, our team has been deep in AWS cost optimization audits: → Why over-sized EC2 instances silently drain budgets → How non-prod workloads running 24×7 inflate AWS bills → Why unused storage and forgotten resources are real cost killers → How teams lose cost visibility as cloud environments scale The insight? Most organizations can reduce 20–40% of AWS spend without impacting performance. Cloud costs don’t explode overnight - they leak quietly until finance notices. If your AWS bill feels “higher than expected, it probably is. 👉 Book a Free AWS Cost Assessment: https://lnkd.in/daHWdtdz #AWS #FinOps #AWSManagedServices #CloudOptimization #CIO #CTO #India #USA #UAE #Europe
-
Terramate
Terraform modules were built for code reuse. They were never meant to be the interface between platform teams and the rest of engineering. At scale — multiple teams, repos, environments — the gaps show up quickly: no enforced directory structure, no built-in promotion model, no visibility into module usage, no governance boundary. Terramate Bundles add the missing contract layer above modules. Platform teams define how infrastructure should be provisioned — naming, security defaults, provider versions, directory layout, lifecycle rules. Developers provision through that contract without touching the underlying Terraform. More in our latest article (see 1st comment). 👇
1 commentaire