About
I am a management graduate with experience of launching v1.0 of a cloud service, scaling…
Activity
4K followers
-
Kotturu Trinadh shared thisGoogle Cloud Storage team collaborated with #PyTorch Lightning AI team to reduce checkpoint times up to 95% using Rapid buckets and GCSFS. Learn more about upstreaming work @ https://lnkd.in/dwwCZQsW Thank you Amit Ponchick, Neil Bhatt, Thomas Chaton, Frances Fedoriska, Liana Koleva, Teja Pulagam from Lightning AI and Ankita Luthra, Jatin Sharma, Zhixiang Li, Mahalaxmi Bejugam, Pavan Daga and Nivedita Aggarwal from GCS team.PyTorch Lightning checkpoints write up to 95% faster on Google Cloud Storage Rapid BucketsPyTorch Lightning checkpoints write up to 95% faster on Google Cloud Storage Rapid Buckets
-
Kotturu Trinadh shared thisAnkita Luthra and I will be talking on how GCSFS on Rapid Buckets helps you accelerate AI model training and inference with high-performance object storage and E2E stack improvements at #Pytorch Conference, North America, 2026.Kotturu Trinadh shared thisvLLM sessions at PyTorch Conference North America 2026 show how teams are optimizing inference, working on PyTorch release validation and compatibility, and using vLLM in broader serving and application workflows. Speakers will cover CUTLASS, Triton, and Helion; faster model loading with fastsafetensors; quantization and HiFloat; autonomous kernel bring-up and Intel GPU optimization; PyTorch release validation, enterprise agentic inference, stable ABI, and dynamic shapes; vision-language serving; cloud-to-edge workflows; high-speed storage; a serving stack for a physics-constrained generative model; and contributor onboarding and first-PR workflows. Featured speakers include: NVIDIA: Michael G., Güray ÖZEN, Aastha Jhunjhunwala, and Mark Moyou, PhD Meta: Andrey Talman, Jane Xu, and Laith Sakka Arm: Kavya Sri Chennoju IBM: Takeshi Yoshimura and Nili Guy (Ifergan) Red Hat: Markell Rawls, Joseph Groenenboom, Tyler Michael Smith, Sean McGovern, Christopher Leonard, and Maroon Ayoub Google: Ankita Luthra and Kotturu Trinadh Intel: Alex Gu, Qun Yang, Wai Hung (Whitney) Tsang, and Artur Fierka Huawei: Jun Zhao and Haonan Zhang IBM Research: Burkhard Ringlein University of Minnesota: Arun Sharma Register for PyTorch Conference North America 2026: https://hubs.la/Q04tBgv_0 Read the vLLM session guide: https://lnkd.in/gAzvw6KQ
-
Kotturu Trinadh shared thisWith the improvements to GCSFS prefetcher, which is made default in 2026.8.0, we are now saturating the NIC of the node. Learn how #Google Cloud Rapid Buckets accelerate the data loading in #PyTorch ecosystem via GCSFS: https://lnkd.in/d-r-J5EJ and https://lnkd.in/dg2qGqT4 Credit to Margubur Rahman for the awesome work that got mentioned in the 10-year retrospective journey of FSspec by its maintainer Martin Durant . Ankita Luthra Jatin Sharma Mahalaxmi Bejugam Zhixiang Li Pavan Daga Nivedita AggarwalGoogle Cloud latest news and announcements | Google Cloud BlogGoogle Cloud latest news and announcements | Google Cloud Blog
-
Kotturu Trinadh reposted thisKotturu Trinadh reposted thisYouTube-scale. Gmail-scale. Gemini-scale. Colossus is the distributed storage system powering Google's largest global services. As the scale and performance demands of agentic-era workloads push the boundaries of storage infrastructure, it made perfect sense to also bring the power of Colossus to Google Cloud Filestore, our NFS file service. So we did that. Backed by Colossus, Filestore is delivering the simplicity, security, and scalability that your workloads deserve. 💪 https://lnkd.in/g8hYaMD5Filestore file service runs on Colossus | Google Cloud BlogFilestore file service runs on Colossus | Google Cloud Blog
-
Kotturu Trinadh reposted thisKotturu Trinadh reposted thisIn the agentic era, AI is evolving from answering questions to reasoning and taking action. Companies who want to lead in this next phase of AI need a computing infrastructure that is designed for these new requirements. So, today, I'm excited to share that Google has just been recognized as a Leader in the very first Gartner Magic Quadrant for AI Infrastructure. We were positioned the highest for 'Ability to Execute' and furthest for 'Completeness of Vision'. Congrats to all the Google teams that deliver on these amazing products, and enable our customers to innovate faster and deliver compelling experiences for their users! Learn more → https://goo.gle/44fBTULGoogle is a Leader in Gartner® Magic Quadrant for AI Infra | Google Cloud BlogGoogle is a Leader in Gartner® Magic Quadrant for AI Infra | Google Cloud Blog
-
Kotturu Trinadh shared thisThis is a wonderful article on World Models in simplest terms by Fei-Fei Li. The beauty of the article is - it explains what a World Model is, without any AI jargon.
-
Kotturu Trinadh shared thisDidn’t knew there was an Agent AI foundation like PyTorch foundation and Linux Foundation.Kotturu Trinadh shared thisAgentic AI is moving from pilots to production faster than any prior enterprise tech wave. The question is no longer whether this ecosystem will be built on open foundations. The question is who shows up to build it. Six months into the Agentic AI Foundation, here's where we stand — the wins, the projects, the community, and the work we have not yet solved. It is an absolute thrill to be working alongside my Linux Foundation colleagues Manik Surtani Angie Jones Alex Salkever Demetrios Brinkmann Jim Zemlin and many other leaders. Together with an extraordinary community of innovators and builders, we are shaping the future of Agentic AI. #AgenticAI #OpenSource #LinuxFoundation #AAIF #MCP #AIagents #FutureOfWork #ArtificialIntelligence #AISix Months of Open: Reflections on the Agentic AI Foundation's First Half-YearSix Months of Open: Reflections on the Agentic AI Foundation's First Half-Year
-
Kotturu Trinadh reposted thisKotturu Trinadh reposted thisIt’s a thrilling time for infrastructure. Reflecting on Gemini 3.5 Flash and our teams’ hard work, I’m excited about our progress. Over the next 8 days, we’ll share a daily digest, ‘8 Days of TPU 8.’ To kick off the series, I want to ground us in four shifts I am seeing across the industry and at Google. Hardware-Software Co-design: System efficiency will become pervasive in public discourse, because designing hardware and software in isolation has hit a wall of diminishing returns. Our work with Google DeepMind on silicon-software co-design allows TPU 8t to achieve nearly 3x higher compute performance per pod than previous versions. Making agentic scale economically viable: Enterprises are transitioning from large training runs to orchestrating agentic workflows, shifting architectural needs and spending to make them economically viable. Real-time, multi-turn agent interactions cause memory bottlenecks; we addressed this in TPU 8i with 3x more on-chip SRAM and 288GB of HBM to host KV Caches entirely on-silicon. And because agents can spend a large part of their runtime sitting idle, we will see a resurgence in CPU demand. Highly fluid general-purpose compute like our Axion CPUs will be vital to handle background tasks and sandboxing without compromising TCO. Designing flexible, reliable and predictive orchestrators: Agentic apps require orchestrators to transition from reactive to predictive, learning-based scheduling. Our experience with Borg and GKE highlights that manual placement heuristics fail under complex workloads. By integrating machine learning to predict task timing and arrival, we can optimize scheduling. We are implementing these strategies via the GKE Agent Sandbox and open-source Agent Substrate. Moving beyond perf/TCO: Jevons Paradox teaches us that a massive increase in the efficiency of a fundamental resource paradoxically leads to an exponential surge in its total consumption. Lowering the cost of cognition will only induce greater demand. This means that as an industry, we must look beyond TCO and engineer our systems for metrics like goodput, intelligence-per-watt and carbon efficiency to deliver increasing value at massive scale. Follow along with Google Cloud #8daysofTPU8 for more technical details and specific engineering choices powering this next frontier and catch the recording from our recent Cloud Next session on “the Future of AI Infrastructure”. https://lnkd.in/ga-dMmkm
-
Kotturu Trinadh posted thisThis is a very interesting article that compares two types of agentic systems: agent teams vs agent swarms. I thought both of them are same until I read this to understand philosophy and use cases: https://lnkd.in/g_72cvu2
Kotturu Trinadh commented on a post
1d
It was great working with you on accessibility for Azure Backup. You can be proud that you made a world a better place for everyone and all the best for your next journey in Europe.
-
Kotturu Trinadh liked thisKotturu Trinadh liked thisCurious how we use agents at Hugging Face ? I explained it all publicly, here are the slides :) The most important aspects: * Personal agents, from the most common tools you will see at HF to the least common * Team agents, our great Team AI everyone uses at HF in personal or team chats * Data Workflows, or where our Team AI gets its superpowers and the security layer being them I got so many good questions on HF Buckets and Xet as well ! Looks like many people about storage deduplication and how it works. Very nice event with Databricks, we talked about AI and Open Source tools for Data in Paris Thanks Carly Akerly Lisa N. Cao Robert Pack Scott Haines Edwin He Aravind Segu for the invitation and great talks. #ai #agents #data #llm #oss #pi #kimi #buckets #parquet #duckdb
-
Kotturu Trinadh liked thisKotturu Trinadh liked thisLots of discussion out there about our next model, so I wanted to give an early look as soon as possible. Introducing Gemini 4 Argon! It shows frontier performance in complex workflows, cyber defense and software engineering. Teams are using it extensively at Google, from coding to quantum computing, great feedback. Importantly Argon has frontier safeguards and we are rolling it out responsibly - it’s with the US gov’t and going to a set of trusted cyber defenders through our Fairwind Program today. We’re going to make it available as soon as we can and as safely as we can. So hold tight, lots more coming, and you’re going to see us iterating rapidly. https://lnkd.in/gTiQb7DA
-
Kotturu Trinadh liked thisIgnite 2026 is just around the corner. It’s going to be amazing speaking alongside Mark - Azure CTO, distinguished CISO and a technical fellow; Amit - Resiliency in Azure GPM, and to top that, we will be joined by a strategic banking customer too - to make it all real and share resiliency insights from lived experiences! We look forward to seeing you in our session ! Do register ! #Ignite2026 #resiliency #AzureKotturu Trinadh liked thisResiliency isnt a milestone you hit — its a lifecycle: Start resilient. Get resilient. Stay resilient. Most teams are strong on one of the three. You design well but drift over time. Or you remediate hard after an incident, then quietly regress. Staying resilient is where the real work lives — and where AI is starting to change whats possible. Thats what I'll be digging into at Microsoft Ignite with Mark Russinovich and Surya Sripathi Raju in AI-powered resiliency for modern workloads — how AI supports each stage of that lifecycle, and what it actually takes to build.📅 [Date] | Session [CODE] 🔗 https://lnkd.in/dKsnN_6S in new window) #MSIgnite #Resiliency #AI #SRE #Azure
-
Kotturu Trinadh liked thisKotturu Trinadh liked thisAs more teams scale open source AI on secure, governed infrastructure, Ray has become critical across the full AI lifecycle, from data curation to production serving and reinforcement learning. Ahead of PyTorch Conference North America 2026 in San Jose, we put together a guide highlighting key Ray-focused sessions on the schedule. Featured speakers include: Anyscale / Databricks / University of California, Berkeley: Ion Stoica Anyscale: Eric Tang, Sumanth R Hegde, Joshua Lee, Mengjin Yan Google: Ankita Luthra, Kotturu Trinadh, Jago Macleod LinkedIn: Tao Huang, Tommy Li Pinterest: Gaurav Arora, Shunyao Li, Eric W. Uber: Ke Chen, Peng Zhang, Xandra Zhu University of Minnesota: Arun Sharma Explore engineering takeaways from these technical sessions where you will learn about unified AI orchestration with Kubernetes, elastic training stacks, and scalable RL. Read the full guide and map out your schedule: https://lnkd.in/e373G-Hu Register for PyTorch Conference North America: https://hubs.la/Q04tBgv_0 #PyTorchCon
-
Kotturu Trinadh liked thisKotturu Trinadh liked thisTomorrow marks my last day at Microsoft—and the beginning of an exciting new chapter as I prepare to relocate to Europe. As I reflect on my time here, I’m feeling incredibly grateful for the opportunities, the challenges that helped me grow, and, most of all, the remarkable people I’ve had the privilege of working alongside. Together, we accomplished work I’m genuinely proud of. Thank you to the colleagues, leaders, mentors, and friends who shared their trust, talent, perspectives, and support so generously. You made my time at Microsoft deeply meaningful, and I’ll carry the lessons, relationships, and memories from this experience with me. I’m leaving with tremendous appreciation for Microsoft and great respect for the people and teams who continue to build, innovate, and make an impact every day. This transition brings together two things I’m excited about: building a new life in Europe and finding my next product management opportunity. I’m especially interested in roles where I can contribute to meaningful, ambitious work alongside thoughtful and talented people. If you know of a team or opportunity that might be a good fit, I’d be grateful to connect. While goodbyes are never easy, I’m genuinely excited about the path ahead—and I hope to stay connected with the many wonderful people who made this chapter so special. Thank you for everything. 🙏 #Microsoft #ProductManager #Compliance #Accessibility
-
Kotturu Trinadh liked thisKotturu Trinadh liked thisProud to announce that Storage Intelligence Advisor - a major new GCS feature for storage management – is generally available now! 🔥 🔥 🔥 👉https://lnkd.in/gqrbGrYR Storage management is one of the most challenging problems at scale, especially given the volume of data created by modern training and inference pipelines. The Storage Intelligence team has been hard at work on two major new features: Storage Intelligence Advisor and Enhanced Batch Operations for GCS. Storage Intelligence Advisor provides out-of-the-box visibility with zero configuration required and baselines your project’s activity to detect anomalies as they happen. Enhanced storage batch operations enables serverless remediation across millions or billions of objects without custom code. Leading teams are already putting this into production: Shipt eliminated complex telemetry pipelines to gain instant anomaly detection, while Palo Alto Networks uses batch operations to seamlessly enforce retention policies across billions of objects. These features are a huge improvement in helping GCS customers manage their storage. Huge congrats to the team on this launch! Yatharth Gupta Raja Loganathan Manjul Sahay Abhishek Lal Himanshu Kohli Santhosh Misro
-
Kotturu Trinadh liked thisLove this breakdown from Cohavit Almagor below. In just my first few months here at #Google, it’s been eye-opening to see how deeply this team understands systems engineering at scale. The industry focuses heavily on model breakthroughs, but once you start running fleets of autonomous workers, you simply can't scale sustainably without purpose-built infrastructure for agents. Filestore agent volumes, paired with GKE Agent Substrate, solves that cleanly from first principles. Massive kudos to the #Filestore team, and our #GKE partners for shipping this today! Highly recommend reading Cohavit’s post below—and checking out the blog link for the technical details.Kotturu Trinadh liked thisFrom a startup vision to the #storage backbone for autonomous #AI #agents 🚀 When Google acquired #Elastifile in 2019, our team set out with an ambitious mission: re-imagine high-performance software-defined file storage as a seamless, enterprise-grade cloud service. Over the years, we transformed that foundation into Google Cloud Filestore scaling it across global enterprises to power mission-critical databases, SAP systems, high-performance computing, and stateful GKE workloads. Today marks an exciting new chapter in that journey. We are officially introducing **Filestore agent volumes** a fully managed storage tier purpose-built for scaling AI agent workspaces on #GCP! As organizations shift from simple prompt-based LLMs to massive multi-agent swarms and coding sandboxes, compute without fast, persistent, and isolated file storage hits a wall. Platform teams have been forced to juggle heavy state shuttling, complex local disk lifecycles, and costly idle storage. Co-designed with **Agent Substrate on GKE** and **GKE Agent Sandbox**, Filestore agent volumes solve these challenges from first principles: 🔹 **Millisecond Attach & Sub-Second Resumption:** Dynamically suspend idle sandboxes to cut compute costs, and resume execution instantly when new tasks arrive. 🔹 **Granular Workspace Isolation:** Strict POSIX boundary controls per workspace so agents can safely execute untrusted code and install toolchains. 🔹 **Pay-Per-Use Elasticity:** Pay only for the capacity consumed, with automatic lifecycle tiering when workspaces are idle. 🔹 **Multi-Agent Collaboration (RWX):** Native Read-Write-Many support and POSIX locking, allowing multi-agent swarms to co-author, test, and review project files concurrently without write collisions. Seeing technology that started as a small, passionate startup evolve into a planet-scale pillar powering the next generation of AI agents is an incredible milestone. Huge kudos to Micah Waldman , Maya Gal, Jerome McFarland , and the entire engineering, product, and GKE #partner teams for driving this forward! 👏 📖 Check out the full launch blog post: https://lnkd.in/dUtugGh3 #GoogleCloud #Filestore #AgenticAI #CloudStorage #GKE #AIInfrastructure #EngineeringLeadership #StartupJourney #DistributedSystems #socialmedia #bettertogether Ronen Schwartz Idan Brown Andi Gutmans Christopher Taylor Sameet Agarwal Iftach Ragoler Prashant Desai Jason Messer
-
Kotturu Trinadh liked thisKotturu Trinadh liked thisA huge unlock for Headroom Labs: We can now show tremendous value in savings with Amazon's Kiro agent - around 60% cost savings in some initial benchmarks(!) on our Enterprise plan Kiro works on the principle of credits, and along with Cursor, is one of the few subscription offerings for coding agents. Multiple teams are leveraging it! If you or your team is exploring Kiro, reach out to us at hello@headroomlabs.ai and be ready to be amazed by our offering.
Experience & Education
Licenses & Certifications
Honors & Awards
-
Co-authored a Business case to Ivey League Cases
Ivey Publishing
Co-authored Business case on "Himalaya Drug Company: Repositioning an Herbal Soap" along with Prof Ramesh Kumar, which deals with Consumer Behaviour and Brand re-positioning of a Herbal Soap offering from Himalaya Soaps, India.
-
National Finalist at Google4Entrepreneurs
Google
Our business plan combining cricket experience with social experience went on to present at Google4Entrepreneurs, a national level B Plan competition.
-
India Leadership Summit
Mckinsey
One of the 60 students selected from across all IIM campuses to participate in the event.
Languages
-
English
-
Other similar profiles
Explore more posts
-
Sathwik Prayakarao
At a certain scale, companies outgrow founder centric execution. What they need instead is: –> institutional leadership –> clearer governance –> and separation between operations and exploration Deepinder Goyal's transition at Eternal fits this pattern. The legacy is already visible: –> a category created –>a platform scaled –> a public company stabilized Stepping back from day to day execution doesn’t reduce impact. In many cases, it changes the altitude at which decisions are made.
-
ILYA FEDOTOV
#AI has reached a tipping point in automation and mass job displacement. It’s here now. There’s nothing left to solve from a theoretical or technical perspective — it just needs to be scaled up. Amazon, the second-largest employer in the U.S. with roughly 1.5 million employees, recently had a leaked memo outlining its plan to automate 75% of its workforce. This isn’t the exception — it’s the new rule that will apply to nearly all public and private companies. It's not necessarily that all of these companies are evil, it's just that there will be no way to compete and stay in business if companies do not adapt as they'll be sidelined by their competition if they don't take advantage of this meta trend. Now that the genie is out of the bottle, the transition will happen faster than people expect. If you’re an individual contributor, you must become not only an expert at using AI, but an expert at building, managing, and scaling teams of autonomous agents. It doesn’t matter what your discipline is — if you’re doing your job professionally (not as a hobby), the expectation will soon be that you know how to do this, just as you now know how to use a computer or advanced software. The difference is that this transition won’t take decades — it will unfold over the next few years. Those who adapt will be the ones who remain in the workforce (aside from government and highly regulated industries, which will behave differently). Individuals who adapt will earn far more in this new environment — not less. However, there will be far fewer of them relative to today’s total workforce. Everyone else will depend on government income, which will lead to mass money printing, and the continual devaluation of fiat currencies for which the only solution is hard assets and capital like bitcoin, gold, property or strong equity. We are past the event horizon here. It will continue to accelerate. Should you panic? Pretty much. You should start learning and building like your future depends on it... because it does. Begin reserving a portion of your workday and personal time to upskill. It doesn't matter if a task seems daunting, AI is getting so good that you want to take on ambitious personal projects and push the frontier. Suggest starting to automate anything you do on a daily basis (emails, comms, news, status reports, writing, social media, etc.), and then start to automate everything you do at work. Just start building or find the right partner to build with, even if it feels scary or out of your depth initially. There’s still time — but not much. Don’t get left behind. Turn the negative into the positive. Be proactive. Take the lead. 💡 Inspired by an X thread of someone I highly admire and whose ideas resonate with me a great deal.
9 Comments
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More