iiinova — where your digital self keeps evolving

Hrittik @hrittikhere.bsky.social · 04/07/2026

Every time an LLM pod restarts on k8s without a cache, it re-downloads the entire model from @huggingface . For a 65GB model, that's 20-40 min of idle pod. Per restart Fix: a PVC + a download Job. Cache once, mount everywhere. Read more here (vLLM on GKE): www.hrittikhere.com/posts/model-...

001

Hrittik @hrittikhere.bsky.social · 26/06/2026

On other news: I have been selected for the first-ever @aaif.io Ambassador Cohort, which feels special because it represents something I've wanted to see for a long time: AAIF becoming a dedicated home for this growing ecosystem 🚀

000

Hrittik @hrittikhere.bsky.social · 25/05/2026

This June is going to be one of the busiest months for me, as I’ll be delivering key talks on some of the biggest stages in India! Very excited to be presenting alongside my fellow speakers and sharing ideas that are changing the way we think about compute, scale & capacity 🚀

100

Hrittik @hrittikhere.bsky.social · 21/04/2026

Ever wondered why GPUs are getting so so important? Here's a small visualization 🤔

000

Hrittik @hrittikhere.bsky.social · 01/04/2026

Today, DevRel isn't just about awareness. It's about getting developers started, helping them adopt the product, and keeping them engaged. I recently had the chance to compile all my notes into this short post with action items: Will love thoughts here: www.hrittikhere.com/posts/devrel...

hrittikhere.com

DevRel Activation Isn't Awarness Anymore: A better PLG Approach | Hrittik Roy

In a Product-Led Growth (PLG) world, DevRel isn't just about awareness anymore. It's about getting developers started, helping them adopt the product, and keeping them engaged.

000

Hrittik @hrittikhere.bsky.social · 19/03/2026

www.hrittikhere.com/talks the perfect DevRel portfolio :)

010

Hrittik @hrittikhere.bsky.social · 16/03/2026

Looks very cool! Can you give some examples of what's high-risk travel/event? I don't think that's about confs?

100

Hrittik @hrittikhere.bsky.social · 15/03/2026

Ever booked a window seat only to be blinded by the sun for 10 hours? 🙈 Check out FlightPath! It analyzes sun exposure and UV levels for your specific route, telling you exactly which side of the plane to sit for shade + sunscreen✈️🗺️ Safe at 35k ft! flightpath.hrittikhere.com

001

Hrittik @hrittikhere.bsky.social · 03/03/2026

Thrilled to be heading to Southern California Linux Expo ( @socallinuxexpo.bsky.social) this week! ✈️ 📍 Ballroom DE 🗓 Sunday, March 8, 2026 ⏰ 11:45 AM – 12:45 PM If you’re attending, drop a comment or DM, always happy to connect in person 🙌

000

Hrittik @hrittikhere.bsky.social · 12/01/2026

Recently, I built my personal portfolio ✨ I wanted it to reflect how I actually learn, share, and engage with the community, rather than just a static resume www.hrittikhere.com

hrittikhere.com

Hrittik Roy | Platform Advocate & CNCF Ambassador

Platform Advocate at vCluster and CNCF Ambassador, working at the intersection of cloud native platforms, open source, and developer experience.

010

Hrittik @hrittikhere.bsky.social · 07/01/2026

🗓 14th Jan | ⏰ 8:30 – 9:30 PM (IST) Join me in conversation with Sagar, our popular @docker.com Captain, for an informal yet insightful session on Docker Hardened Images www.youtube.com/live/tdxUgO5...

youtube.com

The Docker Captain Gossip: Docker Hardened Image with Sagar

YouTube video by Hrittik Explains

000

Hrittik @hrittikhere.bsky.social · 01/01/2026

From HP’s garage beginnings (87 yrs ago) to infinite AI agent loops, self-moving objects, and hard truths about dev jobs this year starts fast 🚀 open.substack.com/pub/hrittikh...

open.substack.com

January 1: "Self-moving mugs," infinite AI loops, and ChatGPT ads?! 🤯 New Year's AI insights are packed: developers' jobs, AI's core tensions, and

Welcome to the January 1, 2026 Edition →

000

Hrittik @hrittikhere.bsky.social · 29/12/2025

December 29: Is the AI bubble about to burst, or is it *really* here to stay? 🤔 We're talking AI agents with narrative identity, China's new rules open.substack.com/pub/hrittikh...

open.substack.com

December 29: Is the AI bubble about to burst, or is it *really* here to stay? 🤔 We're talking AI agents with narrative identity, China's new rules

Welcome to the December 29, 2025 Edition →

000

Hrittik @hrittikhere.bsky.social · 18/12/2025

December 18: AI just got a HUGE upgrade! 🚀 Faster image gen, real-time virtual worlds, & LLMs reaching new heights. Plus, a look at big tech's AI man open.substack.com/pub/hrittikh...

open.substack.com

December 18: AI just got a HUGE upgrade! 🚀 Faster image gen, real-time virtual worlds, & LLMs reaching new heights. Plus, a look at big tech's AI man

Welcome to the December 18, 2025 Edition →

000

Hrittik @hrittikhere.bsky.social · 17/12/2025

Me and @shivaylamba.bsky.social are sitting this Friday to break down one of the most talked-about additions in the Docker, Inc ecosystem: Docker MCP Catalog! Join us live here on 19th December at 7PM IST: www.youtube.com/live/uBdtC3r...

youtube.com

The Docker Captain Gossip: Docker MCP Catalog with Shivay Lamba

YouTube video by Hrittik Explains

010

Hrittik @hrittikhere.bsky.social · 03/12/2025

AWS re:Invent's AI bonanza, Sam Altman's "Code Red," & Home Assistant's local rebellion! Plus, academic drama & AI's wild side. What's *really* happening in tech? Find out! 👇 #AI #TechNews #reInvent #Newsletter hrittikhere.substack.com/p/december-3...

000

Hrittik @hrittikhere.bsky.social · 01/12/2025

there were too many to remember :)))

010

Hrittik @hrittikhere.bsky.social · 01/12/2025

Launched the very first edition! 🥳 We’re covering AI writing its own homework (peer reviews), Black Friday records, and how to train models without melting your GPU. Get the full scoop and subscribe to the daily drop here: hrittikhere.substack.com/p/december-1...

hrittikhere.substack.com

December 1st: AI Peer Reviews, Claude Code, Tim Sweeney’s Take

Welcome to the December 1st or the first Edition → On this day in 1941, Federico Faggin, the Italian physicist and engineer behind the first commercial microprocessor, was born, 84 years ago.

000

Hrittik @hrittikhere.bsky.social · 01/12/2025

There were items like Sprite and local drinks from Costa Rica, so completely different. There was a drink from China that tasted like bacon so a risky adventure

120

Hrittik @hrittikhere.bsky.social · 13/09/2025

🚀 Time flies! I recently wrapped up 𝘁𝘄𝗼 𝘆𝗲𝗮𝗿𝘀 at the @vcluster rocketship More: www.linkedin.com/posts/hritti...

000

Hrittik @hrittikhere.bsky.social · 10/09/2025

🚀 Just wrote about our latest release - vCluster Auto Nodes and how it delivers dynamic #Kubernetes scaling across any infrastructure! Powered by ⚙️ #Karpenter, it gives auto scaling across any infra. Read more in my post: www.vcluster.com/blog/kuberne...

020

Hrittik @hrittikhere.bsky.social · 27/08/2025

Just got my talk in at Cloud Native Rejekts NA (Atlanta) 2025 ⭐

010

Hrittik @hrittikhere.bsky.social · 19/08/2025

I’m thrilled to share that I had the privilege of serving as a Program Committee Member for KubeCon + CloudNativeCon North America 2025, organized by The Linux Foundation! 🎉 #KubeCon #CloudNativeCon

010

Hrittik @hrittikhere.bsky.social · 25/06/2025

Very excited to host Cloud & Compute - AI in Bangalore on July 27, 2025, with Kunal Das and @HowDevelop . A Premier Tech Meetup for Cloud Enthusiasts, AI Practitioners & Infrastructure Experts! We're accepting CFPs so submit now 🥳 sessionize.com/cloud-ai-blr/

sessionize.com

Cloud & Compute - AI: Call for Speakers

Cloud & Compute - AI is an exciting event hosted by Couchbase, LoftLabs and Cloud Computing Circle, bringing together cloud enthusiasts, AI practition...

000

Hrittik @hrittikhere.bsky.social · 23/06/2025

The key insight? If you miss the foundation phase, large communities become glorified help desks. Your early adopters are your future evangelists, so invest in them first. What's been your experience building developer communities? Drop your thoughts below! 👇

000

Hrittik @hrittikhere.bsky.social · 23/06/2025

𝗙𝗼𝗿 𝗟𝗮𝗿𝗴𝗲 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝘁𝗶𝗲𝘀: 1. Avoid the "support center trap" where engagement dies 2. Balance support queries with continuous engagement 3. Leverage demand gen opportunities while maintaining community spirit 4. Scale content efforts strategically (freelancers, agencies, champions)

100

Hrittik @hrittikhere.bsky.social · 23/06/2025

1. Focus on trust and genuine belonging 2. Create easy access (Slack links with clear CTAs on page) 3. Invest in community hours, swag, and appreciation 4. Build your initial champions who become your growth engine 5. Establish regular meet-and-greets with actionable content

100

Hrittik @hrittikhere.bsky.social · 23/06/2025

After working on both sides of developer communities as a member and as a DevRel/Community Engineer, I've learned that 𝘃𝗮𝗹𝘂𝗲 𝗮𝗱𝗱𝗶𝘁𝗶𝗼𝗻 𝗶𝘀 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴, but the approach changes dramatically with scale. 𝗙𝗼𝗿 𝗦𝗺𝗮𝗹𝗹 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝘁𝗶𝗲𝘀:

100

Hrittik @hrittikhere.bsky.social · 19/06/2025

Just finished registering for KubeCon India 🇮🇳. Who am I meeting there? I'm very excited for my second KubeCon! 😁

000

Hrittik @hrittikhere.bsky.social · 11/06/2025

Unlock new levels of networking flexibility & security with vCluster v0.25 and Istio Integration! 🚀 Simplified setup, centralized management, and robust scalability for multi-tenant environments. Learn more in my latest post 👨‍🍳

010

Hrittik @hrittikhere.bsky.social · 03/06/2025

Very excited to share that I’ll be speaking at ‘Infra in Cloud’ this June 7, where I’ll be diving into “Bare Metal Kubernetes: The Past and the Future" :D Looking forward to connecting with the fantastic community at Cloud Native Gurugram. See you there! 🚀

010

Hrittik @hrittikhere.bsky.social · 26/05/2025

🎉 Just passed KCSA (Kubernetes & Cloud Native Security Associate) with 88/100! With focus on Container security, K8s, compliance frameworks, threat modeling it's perfect entry-level cert! Check my repo for tips and areas: github.com/hrittikhere/... #KCSA #Kubernetes

040

Hrittik @hrittikhere.bsky.social · 21/05/2025

Learn event-driven autoscaling in k8s using KEDA’s CPU trigger with GKE! In this step-by-step tutorial, you’ll: - Configure KEDA to scale your deployments based on CPU usage - Simulate traffic to your application and watch KEDA automatically scale your pods real time youtu.be/c4KuDXK0DJ8?...

youtu.be

KEDA CPU Autoscaling Tutorial: Deployments, Load Balancer & Real-Time Scaling Demo

YouTube video by Hrittik Explains

000

Hrittik @hrittikhere.bsky.social · 06/05/2025

Features like multi-cluster, multi-network, and VM support are still in development (track progress on GitHub issues like #54245), but my take is that Ambient Mode’s lightweight design and cost efficiency make it a game-changer for scaling clusters without breaking the bank.

000

Hrittik @hrittikhere.bsky.social · 06/05/2025

The Benefits? → Less Resource Consumption and Cost Savings → Reduced disruption as upgrades can be selectively rolled out without affecting the entire application → Configure L7/L4 as needed according to requirements with a modular architecture.

100

Hrittik @hrittikhere.bsky.social · 06/05/2025

So by breaking them you can choose your components and remove the entire overhead of sidecars. For example, just required mTLS? Use Ztunnel and not the entire stack, while using up to 73% less CPU than sidecar mode.

100

Hrittik @hrittikhere.bsky.social · 06/05/2025

→ Waypoint Proxy: Basically Envoy proxy deployed at the namespace level that enables advanced traffic management (l7) features when needed. Acts as a communication channel between Ztunnel when enabled with the help of Ambient Mode.

100

Hrittik @hrittikhere.bsky.social · 06/05/2025

→ Ztunnel (Zero-Trust Tunnel): For L4 Operations that runs on each node, creating a secure mTLS overlay between different nodes via itself.

100

Hrittik @hrittikhere.bsky.social · 06/05/2025

→ Operational complexity is another pain point. Upgrading Istio requires restarting every application pod to update the sidecar version so your application would have been down while upgrading the mesh Ambient mode introduces a split architecture that breaks the sidecar into two distinct component

100

Hrittik @hrittikhere.bsky.social · 06/05/2025

→ The main challenge? Resource overhead. With one sidecar per pod, a cluster with 1,000 pods requires 1,000 sidecars consuming CPU and RAM. By default, Istio recommends allocating 0.1 CPU cores and 128MB of memory per sidecar.

100

Hrittik @hrittikhere.bsky.social · 06/05/2025

In Istio's traditional sidecar architecture, each application pod runs alongside an Envoy proxy that handles all network traffic. This battle-tested approach has been around since Istio's first release in 2017.

100

Hrittik @hrittikhere.bsky.social · 06/05/2025

Last year in November, Istio’s Ambient Mode reached General Availability with 1.24. But do you know how it compares with the Sidecar mode and what’s better for you and how the architecture compare? Understanding Istio SideCar vs Ambient Mode is essential to unlock the value.

100

Hrittik @hrittikhere.bsky.social · 05/05/2025

→ When the Pod that relies on the secret is removed, kubelet will also eliminate its local copy of the secret data including the linkage and the secret data. Follow @hrittikhere.bsky.social for more such posts! :D

000

Hrittik @hrittikhere.bsky.social · 05/05/2025

→ When the Pod requires it, so the linkage happens through secretKeyRef either via volume or environment variables → Kubelet stores the secret in a tmpfs so that the secret is not written to disk storage, as shown in the diagram. (3/5)

100