Sign in to view Christian’s full profile
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Sign in to view Christian’s full profile
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Phoenix, Arizona, United States
Sign in to view Christian’s full profile
Christian can introduce you to 10+ people at solo.io
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
13K followers
500+ connections
Sign in to view Christian’s full profile
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Christian
Christian can introduce you to 10+ people at solo.io
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Christian
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Sign in to view Christian’s full profile
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
About
Welcome back
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
New to LinkedIn? Join now
Articles by Christian
-
Connecting SaaS MCP Servers to Enterprise With Agentgateway
Connecting SaaS MCP Servers to Enterprise With Agentgateway
Enterprise adoption of MCP still has problems. How do you govern its usage? Especially when developers are willy-nilly…
98
9 Comments -
Enterprise MCP SSO With Microsoft Entra and AgentgatewayJan 26, 2026
Enterprise MCP SSO With Microsoft Entra and Agentgateway
MCP servers are cropping up all over the enterprise like weeds in a nice lawn. And just like weeds, this can cause…
109
8 Comments -
Explaining OAuth Delegation, 'On Behalf Of', and Agent Identity for AI AgentsDec 15, 2025
Explaining OAuth Delegation, 'On Behalf Of', and Agent Identity for AI Agents
As I work with enterprise users adopting AI agents, questions around authorization, impersonation, and delegation come…
145
19 Comments -
Mitigate Prompt Injection Attacks With A2AS and AgentgatewayOct 8, 2025
Mitigate Prompt Injection Attacks With A2AS and Agentgateway
Prompt injection remains one of the biggest open security challenges for AI and LLM-powered systems in the enterprise…
64
7 Comments -
Building an MCP Gateway with Apigee API GatewaySep 30, 2025
Building an MCP Gateway with Apigee API Gateway
Organizations are working out how best to introduce implementations of the model context protocol (MCP) for their AI…
109
20 Comments -
MCP Authorization Patterns for Upstream API CallsSep 17, 2025
MCP Authorization Patterns for Upstream API Calls
The Model Context Protocol (MCP) is moving fast from experimental to enterprise-ready. I am working with a number of…
109
4 Comments -
Authenticating MCP OAuth Clients With SPIFFE and SPIREAug 4, 2025
Authenticating MCP OAuth Clients With SPIFFE and SPIRE
In the previous blog, we dug into dynamically registering OAuth clients leveraging SPIFFE and SPIRE. We used SPIRE to…
147
19 Comments -
Implementing MCP Dynamic Client Registration With SPIFFE and KeycloakJul 30, 2025
Implementing MCP Dynamic Client Registration With SPIFFE and Keycloak
The MCP Authorization spec recommends using OAuth Dynamic Client Registration (DCR) for registering MCP clients with…
164
24 Comments -
Configuring A2A OAuth User DelegationJul 29, 2025
Configuring A2A OAuth User Delegation
In this blog post, we’ll walk through an OAuth 2.0 token exchange and delegation to an A2A Agent.
105
11 Comments -
Enterprise Challenges With MCP AdoptionJul 22, 2025
Enterprise Challenges With MCP Adoption
We know building MCP servers are where everyone’s mind is when it comes to AI agents. That is, if you’re going to build…
236
21 Comments
Activity
13K followers
-
Christian Posta shared thisWe have updated our book "AI Agents in Kubernetes" and gave out about 300 signed copies at KubeCon EU last week! If you couldn't be in person, I invite you to download the digital version (see link in comments). cc Lin Sun
-
Christian Posta reposted thisChristian Posta reposted thisIn her KubeCon keynote, Solo.io Founder & CEO Idit Levine addressed why most AI agent pilots fail to reach production. Developers build fast across frameworks, but Kubernetes lacks native support for agents, tools, and skills. core message: Kubernetes must evolve from cloud-native to agent-native for real production success. https://lnkd.in/eZ8cnCvpSponsored Keynote: From Pilot to Production: Scaling and Optimizing Agen... Idit Levine & Keith BaboSponsored Keynote: From Pilot to Production: Scaling and Optimizing Agen... Idit Levine & Keith Babo
-
Christian Posta reposted thisChristian Posta reposted thisI got to sit down with Louis Ryan and talk about the evolution of service communication, including some of the work we did together on OpenAPI, gRPC, and Service Mesh. https://lnkd.in/emNUwQayThe Evolution of Service Communication: REST, gRPC, and Service Mesh | Louis Ryan & Dan CiruliThe Evolution of Service Communication: REST, gRPC, and Service Mesh | Louis Ryan & Dan Ciruli
-
Christian Posta shared thisI blogged about “agent mesh” a year ago or so. Here it is in action.Christian Posta shared thisThe agentgateway project is best known as a proxy for LLMs, MCPs, and agents. But John Howard and I've been helping organizations use it as an Istio ambient waypoint / egress gateway to solve a surprisingly hard problem: Securely routing pod traffic through a corporate web proxy... transparently. #agentgateway can act as a forward proxy inside the mesh for all traffic (not just agentic): - Pods just call external hosts normally -- no HTTP_PROXY env vars - Proper CONNECT tunnels to the corporate proxy - Auto-injected Proxy-Authorization headers - TLS origination, SPIFFE identity logging, declarative whitelisting If you're dealing with egress + corporate proxy challenges in #Istio, DM me -- happy to walk through the architecture. Check out and ⭐ https://lnkd.in/eu5E7YQk
-
Christian Posta posted thisLove how conference CFPs for AI related/adjacent conferences want you to submit 6 to 8 months in advance. What world is that?
-
Christian Posta posted this🧐 I have been asked a number of times "can i just use OAuth client_id" as my agent identity? The answer I give is "no, it's not a good approach". client_id on its own doesn't strongly prove which specific agent made and executed on a decision. client_id in OAuth is tied to a redirect / callback URL. Anything that has/controls that redirect "is the agent" in that case. 👉 Now if you introduce a client_secret, this a step in the right direction, but still not ideal for AI agents. First, anything that holds that client secret "is the agent". That could be by design (the same type of all gets the same secret), or because the secret got leaked in an env file. 👉 Because of this, enterprises are best to stay away from anything like anonymous dynamic client registration for any type of agent identity, because it will mean essentially nothing. What about Client ID Metadata Documents? ⏲️ This alone is not enough, but again, a step in the right direction. Proving you control a domain and can publish a doc is a stronger guarantee than just bearer credentials, but we can still do better. 🔎 The answer here is tying an agent identity to cryptographic proof. For example something like SPIFFE encodes an agent's identity into a SPIFFE SVID (x509 cert for mTLS or a JWT). 🔎 Another approach is with Agent-Auth (AAuth) that uses keys to sign and verify messages (or delegated tokens issued and backed by dpop keys).
-
Christian Posta shared this🔥 🔥 🔥 Here are my slides from KubeCon EU talk "Enterprise Challenges with MCP Adoption"! Thanks to all who joined live. Recording should be available soon on the KubeCon Youtube channel Cloud Native Computing Foundation (CNCF) solo.io
-
Christian Posta posted thisI am very sorry to hear about the LiteLLM supply chain compromise. But incredible how lucky we are it was discovered and revoked in a couple hours!!
-
Christian Posta reposted thisChristian Posta reposted thisChristian Posta and me will be at our booth at 10:30AM after the day 2 keynote, signing 2 books we did together 🎉 Our AI agents in k8s book was a hit today and we had to hold the remaining 1/3 for the book signing tomorrow.😅 #KubeCon #CloudNativeCon Europe
-
Christian Posta liked thisChristian Posta liked thisHow are AI Agents used? That's the insight from the UK's AI Security Institute. The UK AI Security Institute just published one of the most comprehensive empirical analyses of how AI agents are actually being used in the real world. They analyzed 177,000 MCP tools across 19,000+ servers from November 2024 to February 2026. What's cool is it's not surveys or projects, but actual tool creation and download data from the agent ecosystem. ↳ Software development dominates with 67% of all agent tools and 90% of downloads. ↳ Action tools are surging. The share of tools that let agents directly modify external environments, such as edit files, send emails, execute transactions, rose from 27% to 65% of total usage in just 16 months. Agents are shifting from reading the world to changing it, and this is the most critical finding in the report in my opinion. ↳ General-purpose tools are driving the growth. Browser control, computer use, arbitrary code execution, unconstrained environments where agents can do essentially anything a human can do at a keyboard. ↳ 62% of new MCP servers are now AI-co-authored. Up from 6% in January 2025. Claude Code accounts for 69% of AI-assisted server creation. Agents are building the tools for other agents, and tool creation is no longer bottlenecked by human developers. ↳ Mistakes are already happening in production. The paper documents agents deleting live databases and exposing hundreds of thousands of patient records, not because agents were compromised, but because they had tools to take irreversible actions. A key point is that an agent's risk profile isn't determined by its model. It's determined by its tools. An LLM that can recommend a stock is fundamentally different from an agent that can execute the trade. The action space is the attack surface. This is exactly why monitoring needs to extend beyond model outputs to the tool layer, looking to govern tool call behavior and authorization are so critical, as well as capabilities such as runtime detection and response, hard boundaries and intent analysis. This report is great real-world validation of where agents are already headed, and where security needs to be paying attention as agents begin to become pervasive across enterprise environments.
-
Christian Posta liked thisChristian Posta liked thisIn her KubeCon keynote, Solo.io Founder & CEO Idit Levine addressed why most AI agent pilots fail to reach production. Developers build fast across frameworks, but Kubernetes lacks native support for agents, tools, and skills. core message: Kubernetes must evolve from cloud-native to agent-native for real production success. https://lnkd.in/eZ8cnCvpSponsored Keynote: From Pilot to Production: Scaling and Optimizing Agen... Idit Levine & Keith BaboSponsored Keynote: From Pilot to Production: Scaling and Optimizing Agen... Idit Levine & Keith Babo
-
Christian Posta liked thisChristian Posta liked thisI got to sit down with Louis Ryan and talk about the evolution of service communication, including some of the work we did together on OpenAPI, gRPC, and Service Mesh. https://lnkd.in/emNUwQayThe Evolution of Service Communication: REST, gRPC, and Service Mesh | Louis Ryan & Dan CiruliThe Evolution of Service Communication: REST, gRPC, and Service Mesh | Louis Ryan & Dan Ciruli
-
Christian Posta liked thisChristian Posta liked thisAt KubeCon EU, Maia Iyer and I shared how the SPIFFE Identity Provider for Keycloak can simplify and harden the security of applications. Check out the presentation and demo to see it in action for yourself! * Presentation - https://sched.co/2DY1M * Demo - https://lnkd.in/eYW4Chpi #keycloak #SPIFFE #ZeroTrust #KubeCon #CNCF #RedHat #Security
-
Christian Posta liked thisChristian Posta liked thisI am honored to share that I was elected to the Istio Steering Committee as a community representative for the new 2026-2027 term, along with an excellent group of other contributor and community representatives. I am excited to work with the rest of the committee to help drive the Istio project forward over the next year, as Service Mesh technology continues to evolve and grow in the era of Agentic AI. Istio is a graduated Cloud Native Computing Foundation (CNCF) project that has proven sustained momentum, and with the introduction of agentgateway as a dataplane, the project continues to offer novel capabilities for running large-scale zero trust architectures. Working with Istio at solo.io was a turning point in my career; I pivoted hard into the cloud-native world when I saw the value that open-source infrastructure brought to the table. Kubernetes was a given to me, but Istio was the first project that really sold me on the rest of the cloud-native ecosystem as a unified stack. I continue to be excited every day to leverage cloud-native architecture to solve business problems in my role at GEICO. Faseela K, Craig Box, and zhonghu xu: congrats on your wins! I'm excited to work with you. #istio
Experience & Education
-
solo.io
*** ****** ***** ***
-
*** ****** ******** **********
*********
-
*** ***
****** ********* *********
-
******* ***** **********
******** ******** ******* *********** undefined
-
-
********** ** ****
-
View Christian’s full experience
See their title, tenure and more.
Welcome back
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
New to LinkedIn? Join now
or
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Recommendations received
-
LinkedIn User
“Christian worked for me on a mission critical project that required strong, diverse technical skills and diplomacy. He inherited a C++ application that had evolved over the years from the work of about 5 different people of varying levels of skills and proficiencies. He did an outstanding job ramping up and supporting it. In his tenure he was able to eliminate the reliability issues and inconsistencies in the application and add such effective improvements in functionality and user acceptance that the application was selected to be a corporate application for our agency. Christian is intelligent, professional, diplomatic, and a strong team player. I hope I get the opportunity to work with him again.”
1 person has recommended Christian
Join now to viewView Christian’s full profile
-
See who you know in common
-
Get introduced
-
Contact Christian directly
Other similar profiles
Explore more posts
-
Google AI for Developers
147K followers
Check out these System Instructions for Gemini 3 Pro that improved performance on agentic benchmarks by up to ~5%. For complex agents, enforcing behaviors like persistence, risk assessment, and proactive planning drives better reliability in multi-step workflows. Resources 🔗 SI Template: https://goo.gle/4rBTBMi Docs: https://goo.gle/4oAYN0d
166
9 Comments -
Christian Posta
solo.io • 13K followers
🎯 Running inference workloads on Kubernetes? I just wrote a deep dive on Gateway API Inference Extension and how it works. The Gateway API Inference Extension introduces two new Custom Resource Definitions (CRDs): InferenceModel and InferencePool. By using these two new resources, along with the endpoint selector, your L7 routing infrastructure becomes an “Inference Gateway” enabling you to self-host GenAI/LLMs with a “model-as-a-service” mindset. https://lnkd.in/g_eFqwUd #ai #llm #genai #inference
57
2 Comments -
Hrittik Roy
vCluster • 12K followers
🚀 Bare metal is making a serious comeback, especially for Kubernetes! But when is ditching the VMs for raw hardware really the right move for your K8s clusters? 🤔 It's a hot topic, and many are surprised by the clear advantages in specific scenarios. If you're nodding along, here’s a quick breakdown: 🎯 𝗕𝗮𝗿𝗲 𝗠𝗲𝘁𝗮𝗹 𝗦𝗵𝗶𝗻𝗲𝘀 𝗪𝗵𝗲𝗻: → 𝗠𝗮𝘅𝗶𝗺𝘂𝗺 𝗣𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝗶𝘀 𝗮 𝗠𝘂𝘀𝘁: Direct hardware access means minimal hypervisor overhead and no noisy neighbors hogging resources. Your critical tasks get full infrastructure capacity. → 𝗬𝗼𝘂 𝗡𝗲𝗲𝗱 𝗟𝗼𝘁𝘀 𝗼𝗳 𝗜𝗻𝗳𝗿𝗮 𝗙𝗹𝗲𝘅𝗶𝗯𝗶𝗹𝗶𝘁𝘆: Got specialized, too new, or too old hardware? Bare metal handles it. Access different kernels and fully leverage unique hardware capabilities (like specific network cards) → 𝗬𝗼𝘂 𝗪𝗮𝗻𝘁 𝗙𝘂𝗹𝗹 𝗖𝗼𝗻𝘁𝗿𝗼𝗹 𝗼𝗳 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆: Single tenancy reduces the attack surface. No hypervisor vulnerabilities to worry about, plus you control kernel updates, encryption, and can use hardening frameworks like SELinux/AppArmor. → 𝗬𝗼𝘂 𝗡𝗲𝗲𝗱 𝗠𝗮𝘅𝗶𝗺𝘂𝗺 𝗡𝗲𝘁𝘄𝗼𝗿𝗸 𝗣𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲: Fewer abstractions mean better network speeds. Tools like CNI, flannel, and Calico make setup and routing smoother, and troubleshooting is often simpler. → 𝗬𝗼𝘂 𝗪𝗮𝗻𝘁 𝘁𝗼 𝗖𝗼𝗻𝘁𝗿𝗼𝗹 𝗖𝗼𝘀𝘁𝘀: Optimal resource utilization and predictable scaling costs can lead to significant savings-up to 30% TCO reduction in some cases! Plus, no hypervisor licensing fees. 🚧 𝗕𝘂𝘁, 𝗜𝘁'𝘀 𝗡𝗼𝘁 𝗔𝗹𝘄𝗮𝘆𝘀 𝗦𝗺𝗼𝗼𝘁𝗵 𝗦𝗮𝗶𝗹𝗶𝗻𝗴. 𝗧𝗵𝗲 𝗖𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲𝘀: → 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗖𝗼𝗺𝗽𝗹𝗲𝘅𝗶𝘁𝘆: Forget "click-and-go." Provisioning, updates, networking, DNS, storage, and scaling are all on you. OS patching and certificate management? Your responsibility. → 𝗢𝗽𝘁𝗶𝗼𝗻𝗮𝗹𝗶𝘁𝘆 𝗢𝘃𝗲𝗿𝗹𝗼𝗮𝗱: The "blank slate" can be daunting. Choosing and ensuring compatibility for OS, interfaces, and hardening solutions requires careful planning and dedicated team resources. Choosing bare metal for K8s isn't a decision to take lightly, but for the right workloads, the benefits are undeniable! Learn more in my post here: https://lnkd.in/gqFduUsN
24
-
Jack Vanlightly
Confluent • 3K followers
New blog post: A Fork in the Road: Deciding Kafka's Diskless Future. Kafka is getting serious about S3 and finds itself at an architectural crossroads that will shape its next decade. Several new KIPs (1150, 1176, 1183) aim to reduce replication costs across cloud availability zones, but the implications go far beyond networking cost. It’s a mistake to think of S3 as simply a cheaper disk or a networking cheat. Building on object storage opens the door to operational benefits such as elastic, stateless compute — something many modern analytics systems already exploit. In my latest post, I outline two competing future paths: 🔹 Evolutionary path: Reuse large parts of existing Kafka components to reduce code changes and long-term maintenance. 🔹 Revolutionary path: Separate stateless and stateful layers to realize the full operational benefits of disaggregated storage. The post examines the trade-offs of each path, how the current KIPs map to these two paths and poses a broader question: what should Kafka become? And what will keep it relevant in the decade ahead? https://lnkd.in/d4E2BAe4
149
12 Comments -
Pamela Fox
Microsoft • 13K followers
Our most popular RAG solution now has support for ACLs in the cloud ingestion pipeline! Bring your existing ADLS2 blob storage account or create a new one, and the indexer will extract the oids/groups based off blob ACLs, and store those on the oids/groups fields in the AI Search index. Then, when the user is logged in, they can only chat with documents which they're allowed to access! Release notes here: https://lnkd.in/ewQfJ6Sg
85
1 Comment -
Shannon Williams
Obot AI • 11K followers
Excited to share that we’ve just shipped Obot 0.13, one of our most powerful updates yet. This release introduces composite MCP servers, giving admins the ability to combine multiple servers, expose only the tools they want, and deliver curated experiences to different teams or user groups. We’ve also added audit log export directly from the UI, making compliance, reviews, and long-term record keeping dramatically easier. For teams running Obot on Kubernetes, 0.13 brings centralized scheduling and resource controls, so you can manage affinity, limits, and deployment policies with far more consistency. Admins will now also receive update notifications in the UI, ensuring deployments stay current with the latest improvements. We’ve streamlined the initial setup flow for authentication and model providers, updated documentation across all major clouds, improved resiliency across the platform, and fixed a long list of bugs. A big release with a lot under the hood—excited to see what teams build with it. #AIEngineering #MCPGateway #MCPsecurity #AIops #EnterpriseAI #AIInfrastructure
36
1 Comment -
Stanislav Kozlovski
2 Minute Streaming S.L. • 34K followers
I spoke about just using Postgres instead of Kafka the other day. Here's an Apache-licensed stateless Kafka proxy with a pluggable Postgres (or SQLite, or S3) backend. What stops one from bundling this into Postrges via an extension? It could expose SQL-native functions to use as a pub-sub, and it could expose a true Kafka TCP server (with a background worker). That way it'd remain Kafka wire compatible from within the db machine. And both would share the same tables. I'm not an expert on C nor Postgres extensions, but ChatGPT seems to claim it's doable without much work. https://lnkd.in/dY3Hvwex
87
46 Comments -
Yugabyte
42K followers
"Moving from a single‑node RDS instance to YugabyteDB Aeon has been one of the most impactful infrastructure upgrades in Tofu’s journey. We gained horizontal scale, fault tolerance, and operational simplicity—without sacrificing PostgreSQL familiarity or developer velocity."💬 Don't miss this insightful blog from Ken Kanai, VP of Engineering at Tofu, which details how Tofu evolved its database architecture from a single‑node AWS RDS (PostgreSQL) instance to YugabyteDB Aeon, a fully managed, PostgreSQL‑compatible, distributed SQL database that now powers production.💡 https://lnkd.in/eGBsjHrk #distributedsql #Postgrescompatible #ai #GenAI #database #cloudnative #yugabytedb #opensource #accounting #financialservices
20
1 Comment -
Spectro Cloud
12K followers
Super interesting piece by our partners at TeraSky, introducing their Kubernetes Adoption Framework. KAF is a practical, modular guide to help organizations move from first steps to full production in K8s with confidence. At Spectro, we’re also passionate about simplifying Kubernetes. We know getting the foundation right is critical, and frameworks like this go a long way in clearing the path for others on their K8s journey. Check it out here 👉 https://okt.to/Bt6Xq0 #Kubernetes #KubernetesManagement #CloudComputing #DevOps
22
2 Comments -
MinIO
34K followers
Today marks the launch of Apache Iceberg Digest Vol 1, a new resource built by practitioners for practitioners. https://lnkd.in/gEZhkuAU This first volume dives deep into Iceberg concepts, tools & practical guidance to help engineers and architects stay ahead in their data journey. At MinIO, we believe the debate is settled: Apache Iceberg is the standard for enterprise AI data. With AIStor Tables, we’ve taken the next step by making Iceberg native to the object store. The Iceberg catalog is now part of AIStor itself, not a separate database or service. That means structured and unstructured data finally come together in a single, object-native platform. This ebook showcases the energy and expertise of the Iceberg community, with contributions from leaders across the ecosystem: - Alex Merced, Head of DevRel at Dremio - Karthic Rao, Software Engineer at MinIO - Casey Karst, Principal Product Manager at Fivetran - Kasun Indrasiri Gamage, Product at Confluent - Sa Wang, Software Engineer at PuppyGraph Read it here: https://lnkd.in/gEZhkuAU Apache Iceberg | #ApacheIceberg #AI #DataEngineering
86
2 Comments -
Steve Graves
McObject LLC, embedded… • 4K followers
7 reasons to learn more about eXtremeDB 1) Active Replication Fabric™ to mitigate connectivity issues 2) The only COTS *hard* real-time option 3) Support for shared data in Asymmetric Multiprocessing (AMP) configurations 4) Platform independent interoperability (that’s really 2 but oh well) 5) A type-safe API 6) Record setting time series data processing 7) Pipelining for the lowest possible latency www.McObject.com #extremedb #databasemanagement #iotsolutions #interoperability #EmbeddedSystems #RealTimeData
12
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content