BEGIN:VCALENDAR VERSION:2.0 PRODID:-//HasGeek//NONSGML Funnel//EN DESCRIPTION:Real systems. Real engineers. Real lessons. X-WR-CALDESC:Real systems. Real engineers. Real lessons. NAME:Platform Engineering meet-up - Nov 8 X-WR-CALNAME:Platform Engineering meet-up - Nov 8 REFRESH-INTERVAL;VALUE=DURATION:PT12H SUMMARY:Platform Engineering meet-up - Nov 8 TIMEZONE-ID:Asia/Kolkata X-PUBLISHED-TTL:PT12H X-WR-TIMEZONE:Asia/Kolkata BEGIN:VEVENT SUMMARY:Check-in DTSTART:20251108T091500Z DTEND:20251108T093000Z DTSTAMP:20260421T032822Z UID:session/6Tsq8u6Y4B7iezj1pHeZL6@hasgeek.com SEQUENCE:1 CREATED:20251027T073836Z LAST-MODIFIED:20251027T073840Z LOCATION:Inmobi\, Bangalore ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Check-in in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Introduction to the meet-up DTSTART:20251108T093000Z DTEND:20251108T094000Z DTSTAMP:20260421T032822Z UID:session/13tgSGZwCU3hvpyAHWWGD7@hasgeek.com SEQUENCE:2 CREATED:20251027T073857Z GEO:12.9351;77.6942 LAST-MODIFIED:20251027T075123Z LOCATION:Ground floor Cafeteria - Inmobi\nBengaluru\nIN ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Introduction to the meet-up in Ground floor Cafeteria in 5 min utes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:With great pipelines comes great responsibility: standardizing mul ti-region deployments @Razorpay DTSTART:20251108T094000Z DTEND:20251108T101500Z DTSTAMP:20260421T032822Z UID:session/PqteuTi7fAKEJXJ6sq3dzN@hasgeek.com SEQUENCE:4 CATEGORIES:Talk (30 mins) CREATED:20251027T073939Z DESCRIPTION:### Description\nIn early 2024\, Razorpay’s deployments live d peacefully in a single region - Mumbai. Life was simple until expansion to Singapore began. Suddenly\, our “mostly standardized” Spinnaker pip elines - each with its own flavor of pre-checks\, canary analysis\, and se cret sauce - turned into a spaghetti mess across 150+ pipelines (that too with only a subset of services). Manual edits\, mislinked pipelines\, and a checkout page outage later\, we realized: leaving even a `0.6%` chance o f error in production is still too much chaos for comfort.\n\nThis talk is about how we turned that chaos into order by building **standardized\, po licy-enforced\, and template-driven Spinnaker pipelines**. We’ll walk th rough how we codified deployment safety via **Pipeline-Templates**\, built **custom stages** for Helm releases and multi-stage canaries\, and implem ented **regional isolation** with distributed Spinnakers. Along the way\, we’ll share our lessons from migrating three regions (Mumbai → Singapo re → US)\, introducing MCP (Prompt → Pipeline) tooling for self-serve adoption\, and how this foundational work unlocked larger initiatives like Cell Automation and cross-region deployment safety. Think of it as infras tructure evolution told through the lens of *‘Ramayana for Pipelines’* - where chaos had to cross the ocean to reach production safely.\n\n### * *Takeaways**\n\n1. **Blueprint for Multi-Region Deployment Safety:** Learn how to evolve from ad-hoc pipelines to policy-enforced\, reusable templat es that scale across geographies.\n2. **Balancing Innovation and Control:* * See how to enable safe experimentation (like Mirror Analysis) without le tting chaos slip through production.\n\n### **Who Should Attend**\n\nThis session will resonate with **DevOps engineers\, SREs\, platform teams\,** and **infra architects** who deal with multi-region deployments\, Spinnake r at scale\, or the “joys” of pipeline drift. If you've ever lost slee p over a mislinked pipeline - or want to prevent it before it happens - th is one's for you.\n\n### **Speaker Bio**\n\n**Nanda Kishore** is a Softwar e Engineer at **Razorpay**\, working on large-scale deployment infrastruct ure and release safety across global regions. He has led Razorpay’s tran sition from single-region to multi-region deployments\, building the found ations for Cell-based architecture and standardized release tooling. When not refactoring pipelines or debugging canary rollouts\, he enjoys “vibe -coding” small visual tools to make his life easier - and occasionally\, to keep chaos at bay. GEO:12.9351;77.6942 LAST-MODIFIED:20251027T075111Z LOCATION:Ground floor Cafeteria - Inmobi\nBengaluru\nIN ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/rootconf/platform-engineering-meet-up-nov-8/schedu le/with-great-pipelines-comes-great-responsibility-standardizing-multi-reg ion-deployments-razorpay-PqteuTi7fAKEJXJ6sq3dzN BEGIN:VALARM ACTION:display DESCRIPTION:With great pipelines comes great responsibility: standardizing multi-region deployments @Razorpay in Ground floor Cafeteria in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Q&A and transition DTSTART:20251108T101500Z DTEND:20251108T102500Z DTSTAMP:20260421T032822Z UID:session/WShiFVPRkyChUcwNPNFmi2@hasgeek.com SEQUENCE:3 CREATED:20251027T074031Z GEO:12.9351;77.6942 LAST-MODIFIED:20251027T075118Z LOCATION:Ground floor Cafeteria - Inmobi\nBengaluru\nIN ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Q&A and transition in Ground floor Cafeteria in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Demo - Designing a test observability platform: from smarter build s to AI-driven RCA DTSTART:20251108T102500Z DTEND:20251108T104000Z DTSTAMP:20260421T032822Z UID:session/AxXWHnfhesyCJpwuJHQqu6@hasgeek.com SEQUENCE:6 CATEGORIES:Demo (5-10 mins) CREATED:20251027T074135Z DESCRIPTION:Session Description \n\nIn this session\, we will explore how to design a Test Observability Platform that provides deep visibility into tests\, builds\, and failures across the CI/CD pipeline. The platform com bines data from logs\, traces\, and test reports to deliver actionable ins ights\, enabling teams to detect\, analyze\, and resolve failures much fas ter. We will discuss how selective builds\, coverage-based test selection\ , and shift-left vulnerability scanning fit into this platform to improve developer feedback and overall quality.\n\nThe session will also cover how AI/ML can be applied to observability data for smarter triage and root ca use analysis (RCA). By correlating signals from OpenSearch (logs)\, Jaeger (traces)\, and ReportPortal (test outcomes)\, the platform can classify f ailures\, detect anomalies\, and suggest probable root causes. Attendees w ill learn how to architect such a system\, the trade-offs involved\, and h ow it drives efficiency for engineering and QA teams.\n\nKey Takeaways \n\ nHow to design a Test Observability Platform that unifies build optimizati on\, test execution\, and RCA.\n\nHow to apply AI/ML on observability data (logs\, traces\, reports) for faster and more accurate failure analysis.\ n\nTarget Audience\n\nQA and Test Engineers who want deeper visibility int o test executions.\n\nDevOps/SRE teams working on scalable and reliable CI /CD pipelines.\n\nArchitects and Engineering Managers looking to shift-lef t testing and observability in their organizations.\n\nBio\n\nI am a Princ ipal Engineer at Acceldata\, driving developer productivity through scalab le CI/CD\, observability\, and automation. My work includes building AI-po wered agents for build failure analysis\, designing reliable release and t est infrastructures\, and integrating DevSecOps practices across pipelines . I focus on microservices optimization\, distributed tracing with Jaeger\ , and centralized log management with OpenSearch. Additionally\, I create engineering dashboards for real-time visibility into builds\, deployments\ , and test metrics. My goal is to enable faster feedback cycles\, higher c ode quality\, and secure\, resilient delivery pipelines. GEO:12.9351;77.6942 LAST-MODIFIED:20251107T132709Z LOCATION:Ground floor Cafeteria - Inmobi\nBengaluru\nIN ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/rootconf/platform-engineering-meet-up-nov-8/schedu le/designing-a-test-observability-platform-from-smarter-builds-to-ai-drive n-rca-AxXWHnfhesyCJpwuJHQqu6 BEGIN:VALARM ACTION:display DESCRIPTION:Demo - Designing a test observability platform: from smarter b uilds to AI-driven RCA in Ground floor Cafeteria in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Q&A and transition DTSTART:20251108T104000Z DTEND:20251108T105000Z DTSTAMP:20260421T032822Z UID:session/BvriJGsHgcKTzh62qzu6i@hasgeek.com SEQUENCE:3 CREATED:20251027T074128Z GEO:12.9351;77.6942 LAST-MODIFIED:20251107T132714Z LOCATION:Ground floor Cafeteria - Inmobi\nBengaluru\nIN ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Q&A and transition in Ground floor Cafeteria in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Sponsored talk: Building Low-Latency ML Inference at Ad-serving Sc ale DTSTART:20251108T105000Z DTEND:20251108T112500Z DTSTAMP:20260421T032822Z UID:session/9Je6yfQ2ESvzHWGz9Gr1Xn@hasgeek.com SEQUENCE:8 CATEGORIES:Talk (30 mins) CREATED:20251027T074042Z DESCRIPTION:Ever wondered what it takes to run a low-latency\, high-throug hput ML platform serving over 3 million requests per second? In this talk\ , we’ll dive deep into the engineering challenges of operating at extrem e scale — from handling synchronous ML inference in real-time ad serving to ensuring predictable latency under heavy load. We’ll explore the evo lution of our architecture — how Jetty-based application logic interacts with TensorFlow Serving in a latency-critical path.\n\nWe’ll discuss re al-world lessons on ML inference optimization\, timeout budgeting\, query planner to optimise the I/O calls . You’ll learn how we tuned system con figs\, optimized model serving\, and built guardrails for graceful degrada tion without hurting user experience — all while running in production a t massive scale.\n\nKey Takeaways:\n\nPractical strategies to reduce ML in ference latency in high-QPS environments.\nDesign patterns and system conf igs that help achieve consistent sub-100ms response times under real-world load.\n\nWho Should Attend:\n\nBackend Engineers building latency-sensiti ve or ML-driven systems.\nMLOps & Infra Engineers managing large-scale mod el deployments.\nArchitects designing scalable\, fault-tolerant inference and serving pipelines.\n\nSpeaker Bio\nShivam Gupta\nStaff Engineer @Inmob i \nDSP team\n\n GEO:12.9351;77.6942 LAST-MODIFIED:20251107T132716Z LOCATION:Ground floor Cafeteria - Inmobi\nBengaluru\nIN ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/rootconf/platform-engineering-meet-up-nov-8/schedu le/building-low-latency-ml-inference-at-ad-serving-scale-9Je6yfQ2ESvzHWGz9 Gr1Xn BEGIN:VALARM ACTION:display DESCRIPTION:Sponsored talk: Building Low-Latency ML Inference at Ad-servin g Scale in Ground floor Cafeteria in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Closing notes DTSTART:20251108T112500Z DTEND:20251108T113500Z DTSTAMP:20260421T032822Z UID:session/8aWseHErMtqJYBJfaEVVLW@hasgeek.com SEQUENCE:2 CREATED:20251027T074230Z GEO:12.9351;77.6942 LAST-MODIFIED:20251027T075143Z LOCATION:Ground floor Cafeteria - Inmobi\nBengaluru\nIN ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Closing notes in Ground floor Cafeteria in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Networking DTSTART:20251108T113500Z DTEND:20251108T120000Z DTSTAMP:20260421T032822Z UID:session/yzwam1UQF5Uw9D3v5djuC@hasgeek.com SEQUENCE:2 CREATED:20251027T074239Z GEO:12.9351;77.6942 LAST-MODIFIED:20251027T075148Z LOCATION:Ground floor Cafeteria - Inmobi\nBengaluru\nIN ORGANIZER;CN=Rootconf:MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Networking in Ground floor Cafeteria in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT END:VCALENDAR