Kunal Verma

The new china visa process is really making it easy for them to say things are missing even after details have been shared @ChineseCGMumbai Then they ask you to make the payment again after rejecting it for absurd reasons, this used to never happen when the process was in-person

Felt like writing today, something I was always curious about is @vcluster and today along with @kgatewaydev Article link below

Recording is live, watch now: youtube.com/live/OKPmMpODP…

Kicking off 2026 with the first webinar of the year 🔥 Had a great time hosting a session on - Production @opentelemetry on Kubernetes with @yashsharma_f from @digitalocean . We covered what it actually takes to run OpenTelemetry in production. From infra + app correlation to a live demo with Prometheus, Loki, and Tempo coming together in a unified view with @randoli_inc . The demo ended up being one of the highlights of yesterday’s session, and it definitely set the tone for the upcoming sessions we’ve planned. Big thanks to Yash and everyone who joined. This is just the first of many sessions we’re planning this year 👀 If you’re running OpenTelemetry in production (or planning to), this is a good watch. Check out the recording (link below) 👇

@Rahulktwt @yashsharma_f Thank you joining the session, Rahul. Appreciate it 💙

Hey @yashsharma_f and @kverma_, Thank you for the awesome webinar on OpenTelemetry on Kubernetes. I joined in the middle of the demo and forgot to say Hello, but I truly appreciate the session and the effort you put into delivering such a clear and valuable demo.

Your budget for ingesting logs and traces should be $0. While Dynatrace uses obscure host-unit pricing, we process data locally so you only pay for the insights you actually use. Reclaim your budget today.

More data ≠ better observability. While giving a talk at KCD Bengaluru last year, almost everyone had observability set up. But, almost no one felt it was actually effective. Meaning? High volume, low signal = high cost. This is why @opentelemetry matters: - instrument once, - then control what you send, where, and how. Quality > quantity.

@i_siddhantk @Payoneer Yeah, exactly!

@kverma_ @Payoneer OMG! This is so wrong!

Very disappointed with @Payoneer ! My account was abruptly closed without any prior notice, citing vague “Terms & Conditions” issues, and the email explicitly says it cannot be reactivated and that support won’t respond to queries. This is extremely concerning behavior for a financial services company. I’m simply asking for transparency and a proper explanation. Hoping the Payoneer team can look into this. cc @Payoneer_Help

Last weekend was super fun 🚀 I had the opportunity to speak at CNCG Noida’s recent event called “MissionO11yPossible: Panic at the Pod S2.” It was a fully hands-on event where 12 teams tackled real on-call scenarios, used observability platforms, and performed root-cause analysis under pressure. I gave a live demo of @randoli_inc , talked about OpenTelemetry-native observability, and it was genuinely exciting to see the level of curiosity and enthusiasm around @opentelemetry in the community right now. Huge shoutout to the organizers, judges, and all the participants for making this a memorable event 💙

RSVP now: lnkd.in/dnQuuAnY

⏰ Reminder: We’re going LIVE tomorrow! We’re closing out the year with a conversation we believe many teams should be having right now: “How platform engineering and observability are reshaping developer experience.” If you’re building internal platforms, working on DevEx, or thinking about observability beyond dashboards, this one’s for you. Join us for a candid panel discussion with folks building and running platforms at scale: ➡️ Cortney Nickerson ➡️ Atulpriya Sharma ➡️ @ramiyengar ➡️ @rajith_a Set your reminders now & see you there (link below) 👇

@karanjagtiani04 @linuxfoundation @randoli_inc Appreciate it, Karan!

@kverma_ @linuxfoundation @randoli_inc Nice work, Kunal! Figuring out observability can be tricky-great to see your progress. Good luck with the CKA!

Excited to share that I’ve passed the OpenTelemetry Certified Associate (OTCA) exam by the @linuxfoundation 🎉 When I joined @randoli_inc about a year ago, Observability was a completely new space for me. There were so many tools, concepts, and patterns that initially felt overwhelming. Over the past year, working with @opentelemetry and helping users adopt it has helped me better understand not just the technology, but also why it matters in real systems. Preparing for this exam helped me strengthen my fundamentals and gain a deeper understanding of not just the tool, but how observability works under the hood. Looking forward to applying this learning across different use cases and creating more impact. Grateful to @CloudNativeFdn @CNCFAmbassadors and @linuxfoundation for the support and for building such a great learning ecosystem 💙 Next up - CKA 🙇‍♂️

My friend @xojessfrick organized this for our friend @tessak22 - help if you can. gofund.me/b8769ba80

➡️ The Silent Failures

➡️ Model Context Health

If you’re building GenAI apps and only watching CPU, memory, and error rate, you’re probably missing the real failures. LLM systems don’t fail like traditional systems. Infra can be green. Latency can look perfect. And the output can still be nonsense. And, what’s the part most dashboards won’t tell you. As I’ve been diving deeper into GenAI / LLM observability, one thing has become clear to me: Monitoring infra alone is no longer enough. You need to observe the model too. As more systems become LLM-first, SREs are being forced to rethink what “healthy” even means, because uptime doesn’t guarantee correctness anymore. So, here are the 5 golden signals SREs should rethink for LLM workloads 👇🧵