Omkiran S.

Is our obsession with 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 distracting our focus? In the post ZIRP world where money is no longer free and mass layoffs are prolific we’re increasing seeing organisation's shift towards focussing on “performance”. Not inherently bad in itself. 𝗗𝗼𝗲𝘀 𝗮𝗻𝘆𝗼𝗻𝗲 𝗻𝗼𝘁 𝘄𝗮𝗻𝘁 𝗮 𝗵𝗶𝗴𝗵 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗶𝗻𝗴 𝘁𝗲𝗮𝗺? Naturally talent plays a huge role in this but never has hiring a group of talented individuals made a successful team. Our best people can be easily broken in a bad environment. The obsession with talking about performance needs to stop and replaced with an intent on sculpting the environment for high performance. So how do we create the environment for performance? 🎯 𝗡𝗼𝗿𝗺𝗮𝗹𝗶𝘀𝗲 𝘀𝗮𝘆𝗶𝗻𝗴 “𝗻𝗼” - Inevitably new initiatives or tasks come in from left field that need to be prioritised. As leaders it’s important to ask questions like “what needs to drop to accommodate this ask?”. Check-in regularly with your directs to understand their current WIP (work in progress) to gauge how sustainable their workload is. Make it clear that saying no is not a sign of under-performance but instead a sign of maturity in managing their own workload and being focussed. 🎯𝗖𝗿𝗲𝗮𝘁𝗲 𝗳𝗼𝗰𝘂𝘀 𝗯𝘆 𝗿𝗲𝗱𝘂𝗰𝗶𝗻𝗴 𝗪𝗜𝗣 - I find you get the best out of people when they operate at about 70-80% capacity - it’s certainly how you get the best out of me. This is often incorrectly confused with going slow. Rather we’re more effective when we focus on fewer, but higher priority initiatives. 🎯𝗠𝗮𝗸𝗲 𝗶𝘁 𝘀𝗮𝗳𝗲 𝗳𝗼𝗿 𝗮𝗹𝗹 𝘃𝗼𝗶𝗰𝗲𝘀 𝘁𝗼 𝗯𝗲 𝗵𝗲𝗮𝗿𝗱 - Every leader will say they want psychological safety but that level of trust is earned to the point where the gap between the grapevine and speaking truth to power is near zero. It takes courage, vulnerability and emotional intelligence as a leader to create this environment. The default should be to assume you don’t have it and work tirelessly to build it. What am I missing? What else is critical to building an environment for performance?

32

4 Comments

Istio Service Mesh: great to see Ambient moving up to Beta, delta xDS becoming the default, and api-gateway becoming a first class citizen in Istio. https://lnkd.in/gYPmnTuj

34

🗓️ Today marks six years full time as CTO at SunGod so I pause to reflect and consider six things I wish I knew on my first day back in 2018. 1. Prioritise ruthlessly. You will never finish your task list so identify the items that truly move the needle, both short and long term, and focus on them. Interesting ≠ Impactful. But remember... 2. Your priorities are just your perspective. Communicate with everyone, deeply understand the why behind the what and put that in the context of the business as a whole. 3. Maintenance cost is real. And in a small team, that cost is your time. You'll recognise this when trying to fix a 'one-off' tool you built for a partner many moons ago on a legacy system you're no longer familiar with. The best code is that which never has to be written. 4. Empower selfishly. Hand over the keys, share your knowledge and relinquish control. In return you'll gain time, diverse ideas and a team who GetShitDone™ without relying on you. 5. The best language or framework may not be the best for your team. You may absolutely love Vue.JS and Nuxt but if there are fewer developers is it still the right choice? This extends to all choices so intentionally consider the long term. 6. Celebrate the wins. It is far too easy to immediately focus on the next project, task or problem. Be proud of your work, your team and the mission your building towards. In another six years I'll no doubt have a new list and look back questioningly at this one, but is that not the whole point of growth? #Workiversary #Growth #Learnings

55

2 Comments

Infrastructure scaling at beast scale is a different ball game altogether. Learn from this insightful series penned by Ajay Choudhary on how we rewrote the playbook and how we stumbled upon hard limits of various infrastructure layers - overcoming them one by one. Don't miss this one!

60

A super clear, straight forward example of how to do cost effective RAG using LangChain for your document pipeline and Groq for compute resources. The code demonstrates how to speed up RAG and reduce costs using quantized embeddings (reducing 32 bit values to 8) Nice job Sahibpreet Singh! https://lnkd.in/gf5QMCKK

44

2 Comments

After GECO and Super General adopted Sugar Sell from to improve lead management and customer engagement-enhancing marketing approaches and project sales, they saw: 🚀 26% increase in cases per technician ⚙️ 10% increase in first-time fix rate https://sgrcrm.co/3UpigoE #CRM #SugarCRM

6

I am excited to share the latest blog, co-authored by myself, Junhwi Kim and Vyas Sekar, detailing how we managed to successfully localized and improved LLM by combining Fine-Tuning and RAG. https://lnkd.in/gKDkgGPq

70

3 Comments

By combining mabl and honeycomb.io, you can extend #Observability During Development into your automated end-to-end test runs. In this post I show a couple of techniques using OpenTelemetry to correlate your test runs with your system traces that can turbo-boost the dev-test experience. I wrote this post during the launch of Honeycomb for Frontend Observability. It scratches the surface of pairing mabl's awesome Browser automation with rich frontend traces. ...and I've open-sourced testevents (https://lnkd.in/e76_4n4x) to help create distributed traces. Thanks mabl for publishing this and helping guide the post for your audience.

52

<rant subject = #streaming and #queues> Streams have stronger ordering guarantees than message queues. Generally this means that all messages within a partition are read in the same order. This is highly useful when you need an ordered record of events that happened. As a consequence, streaming systems don't have individual message acknowledgement as it would violate ordering guarantees. Message queues have per-message acknowledgement, which means they have less strict ordering guarantees. One hybrid approach which is quite cool is "key-based subscriptions" in Apache Pulsar and also Confluent's Parallel Consumer for #apachekafka which guarantee order for all messages with the same key while allowing you to individually acknowledge messages. The tradeoff with individual message ack's is persistence (disk or memory)—you need to keep track of which messages have been acknowledged and which ones are still inflight. Pulsar uses BookKeeper to persist "subscriptions." Parallel Consumer stores them in the user-defined metadata in the consumer offset topic. There's an in-flight KIP for "Queues on Kafka" (KIP-932) but if I were the Grumpy Maintainer of Kafka, I would say that it's unnecessary given that it's already possible to implement queue semantics on top of Kafka in the client side (see: Parallel Consumer). Why add additional code bloat on the broker side when the features are already accessible? </rant>

4

Ready to Cook Up Some AI magic....? Come join me for the Culinary Adventure of "𝐓𝐞𝐬𝐭𝐢𝐧𝐠 𝐰𝐢𝐭𝐡 𝐚 𝐃𝐚𝐬𝐡 𝐨𝐟 𝐀𝐈 𝐀𝐰𝐞𝐬𝐨𝐦𝐞𝐧𝐞𝐬𝐬" at the BrowserStack office in Mumbai on August 3rd at 10 am IST. What to Expect: - Delicious Demos of #AI tools in action - Secret Recipes for smarter, faster testing - Tasty Tips to overcome common #testing challenge Register now and bring your appetite for innovation! (Link in the comments) BrowserStack #QAMeetup #Mumbai

67

3 Comments

📢 Exciting News from Western Digital! 📢 Check out this video on the AI Data Cycle to learn more on how Storage and innovation across HDD and Flash is empowering the world to harness the full potential of AI. 🚀 In this video, Western Digital executives dive into the critical role that data storage plays in the AI ecosystem and how our cutting-edge solutions are enabling future growth. From data creation and capture to storage and processing, Western Digital's HDD and Flash technologies are at the forefront, enabling seamless AI integration and optimization. On a personal note, having started in the HDD industry almost 25 years ago working on 10GB HDDs, it's incredibly satisfying and humbling to now be involved in delivering 32TB, 40TB, and even 50TB HDDs. It's satisfying to see the contributions and perseverance pay off, and humbling to collaborate with such amazing innovators and people I genuinely care about. Stay tuned for more throughout the year as we announce more news on our upcoming roadmaps. #AI #DataCycle #Innovation #HDD #Flash #WesternDigital #DataStorage #Technology #FutureOfAI

22

🚀 Curious about how we scale Disney+ Hotstar's infrastructure to handle millions of concurrent users? In my latest blog, I dive into the journey of optimizing core infrastructure components, tackling challenges head-on, and redefining our approach to scalability and reliability. This journey has been truly transformative! 📈 Not only did this transformation improve performance, but it also set the stage for future growth. Check it out and explore the insights that shaped our success! A special thanks to Harsh Mittal Karthik Vijayaraju and Anil Kumar A. for your invaluable contributions, guidance, and support in writing this tech blog. Disney+ Hotstar #infrastructure #engineering #devops #scaling #cloud

108

1 Comment

November 1st marked my 6th year at Whimsical. Going from Leadership → Engineer → Leadership as employee #4 has been quite the journey. Few things are ever clear in startup life, and situations change rapidly. Today, I want to share something that's been instrumental in helping me stay grounded and non-burned-out: working with a Leadership coach! I think we've all felt it—the more senior your role, the less effective your previous support systems are. Your first manager might have been an amazing mentor, but eventually, that tends to change.  And usually - not because they wouldn't want to! It’s the dynamics, the different approaches, different characters that tend to stand in the way of that support being just what you need. Over the years, I've had the opportunity to work with three great coaches, each uniquely valuable. My first coach, an entrepreneur outside tech, made a bold claim: "I can take your anxiety away in 3 months." I mentally called BS. Three months later, my anxiety dropped from 7/10 to 2/10.  It took a lot of trust and vulnerability to open up the way that this particular relationship asked for, but the end results were nothing short of spectacular for me. After a year and a half of transformative sessions, we mutually agreed we'd reached a natural endpoint - an important lesson about recognizing when it's time to move on. My second coach, an ex-CTO from a hypergrowth startup, took a radically different approach. He'd interrupt me with "no no no, that doesn't answer my question" when I tried to dodge uncomfortable truths. Exhausting? Yes. Transformative? Absolutely. When Whimsical growth plans shifted post-Covid, we had an honest conversation about parting ways - teaching me that these endings can be liberating rather than dreaded. My third coach, an ex-military officer, has been another incredibly different set of experiences. Several of our leadership team have, at times, worked with him, creating opportunities for mediated conversations through a trusted third party who had "no horse in the race." The freedom to have difficult conversations without overthinking every word has been invaluable. What has since surprised me most is how often mentioning leadership coaching sparks excited conversations with industry peers. People want to know more and feel validated that it's okay to seek support. Some of my personal key insights: - The right coach can transform your mental well-being and leadership effectiveness - Most senior roles (IC, Manager, C-suite) benefit from having a risk-free space for dialogue - Yesterday's perfect coach might not be tomorrow's perfect coach - Your ideal coach won't be everyone else's ideal coach  I'm really curious to hear from you: - Have you worked with a coach? - How did you find them? - How did you know they were the right fit? I'd love to hear about your experiences! Also, LinkedIn's character limit is trash. I had to cut lots of details. Feel free to ask in the comments.

55

3 Comments

A core tenet for #AWS has always been that, if it was possible to take care of scaling, resilience, availability, security, and cost management of our customer's infrastructure components, we should do so. Eventually, this became known as "serverless", which encompasses much more than executing code without the need for servers. Amazon SQS and S3 were the first AWS services launched, and they were "serverless" from day one. A good example of this philosophy in practice is Aurora Serverless, which provides on-demand auto-scaling for Amazon Aurora databases. It proactively scales up resources during peak periods to ensure the predictable performance when you require it. When demand subsides, it seamlessly scales back down, reducing waste. And with granular, second-by-second billing, you can truly optimize for cost. Remember, the essence of frugal architecture is robust monitoring driving the ability to optimize costs. To dive deeper into how Aurora Serverless exemplifies this principle, I recommend reading the latest paper here: https://lnkd.in/e4uPK6wv #AWS #serverless #databases

1,024

16 Comments

Eliminating dupes with Kyloe DataTools ✨ Manually merging 30,000 duplicates in Bullhorn isn't realistic. So, what can you do? IDR chose Kyloe DataTools to get rid of their dupes, and now have the benefit of being able to automatically merge them before they even know they're there, or do ad-hoc merges when they need to clean things up. Learn more about what Kyloe DataTools can do for you: https://lnkd.in/g4syS_z3 Thanks to Will Hayes at IDR for your feedback! #DataManagement #Recruitment #Dedupe #Duplicates #KyloeDataTools #ThinkBullhornThinkKyloe

2