BEGIN:VCALENDAR VERSION:2.0 PRODID:-//wp-events-plugin.com//7.4.0.1//EN TZID:Asia/Kolkata X-WR-TIMEZONE:Asia/Kolkata BEGIN:VEVENT UID:217@cds.iisc.ac.in DTSTART;TZID=Asia/Kolkata:20260729T160000 DTEND;TZID=Asia/Kolkata:20260729T170000 DTSTAMP:20260723T142256Z URL:https://cds.iisc.ac.in/events/ph-d-thesis-colloquium-102-cds-29-july-2 026-serverless-orchestration-of-classical-and-agentic-workflows-on-multi-c loud-faas-platforms/ SUMMARY:Ph.D: Thesis Colloquium: 102: CDS: 29\, July 2026 “Serverless Orc hestration of Classical and Agentic Workflows on Multi-Cloud FaaS Platform s” DESCRIPTION:DEPARTMENT OF COMPUTATIONAL AND DATA SCIENCES\nPh.D. Thesis Col loquium\n\n\n\nSpeaker : Mr. Varad Kulkarni Vinod\nS.R. Number : 06-18-01- 10-12-21-1-19680\nTitle : “Serverless Orchestration of Classical and Age ntic Workflows on Multi-Cloud FaaS Platforms”\nResearch Supervisor : Pro f. Yogesh Simmhan\nDate &\; Time : July 29\, 2026 (Wednesday)\, 16:00 P M\nVenue : #102\, CDS Seminar Hall\n\n\n\nABSTRACT\nFunctions-as-a-Service (FaaS) has become a natural building block for elastic\, event-driven clo ud applications: developers write short-lived functions\, and the platform handles scaling and billing. As applications grow beyond single functions \, providers expose workflow orchestrators\, such as AWS Step Functions an d Azure Durable Functions\, that compose these functions into multi-step p ipelines. At the same time\, a new class of applications is emerging: agen tic workflows\, in which large language model (LLM) agents iteratively rea son\, call tools\, and adapt their control flow based on intermediate prog ress. Orchestrating both classical and agentic workflows efficiently acros s multiple clouds remains difficult. Provider platforms are proprietary an d asymmetric\; placement and fusion decisions lack principled multi-cloud algorithms\; and the stateful\, iterative nature of agentic patterns sits uneasily with the largely stateless FaaS execution model.\n\nThis talk pre sents a systems stack for serverless orchestration of classical and agenti c workflows on multi-cloud FaaS platforms\, and closes the loop with a liv e application of that stack in graduate education.\n\nWe begin by asking w hat modern FaaS workflow platforms actually cost and where time goes. Thro ugh XFBench\, a cross-cloud benchmark suite spanning AWS and Azure\, we ch aracterize orchestration and inter-function communication as first-class p erformance factors\, and show that cold starts and topology interact in wa ys that single-function benchmarks miss. Building on these asymmetries\, w e design XFaaS: a portable compilation and optimization layer that deploys provider-independent workflows onto native AWS and Azure orchestrators\, and jointly decides where functions should run and which adjacent steps sh ould be fused. The goal is not a single “best cloud\,” but principled hybrid-cloud planning under latency and cost objectives.\n\nThe second hal f of the talk turns to agentic workloads. We introduce FAME\, a middleware that maps common agent patterns onto serverless orchestration with durabl e memory and Model Context Protocol (MCP) tooling\, so iterative LLM agent s can run without reinventing state and tool plumbing for each cloud. We t hen extend portability to multi-cloud agentic settings with XFAGENT\, whic h supports conditional loops and declarative agent configuration while gen erating native orchestration code across providers. Together\, these layer s treat agentic applications as first-class citizens of the FaaS workflow ecosystem rather than ad hoc scripts around chat APIs.\n\nFinally\, we val idate the stack outside micro-benchmarks by deploying it in a live graduat e Cloud Computing course at IISc. An in-class AI Tutor Agent is coupled wi th Moodle-triggered FaaS evaluation workflows that generate student feedba ck and instructor analytics under the bursty submission patterns of real l ectures. Alongside a systems evaluation of these classroom workflows\, we report pedagogical insights from linking tutor transcripts to proctored as sessment: how students engage with the agent\, where transcript coverage a ligns with quiz performance\, and where discrimination among related conce pts still breaks down despite exposure. Beyond showing that the systems st ack is operationally viable for education\, the deployment raises a deeper question for AI-native applications: scalable orchestration is necessary\ , but application-level alignment between dialogue\, analytics\, and asses sment still determines whether the system delivers meaningful outcomes.\n\ nOverall\, the talk argues that classical and agentic multi-cloud FaaS orc hestration can be made portable\, optimizable\, and practically deployable \, and that closing the loop with a real classroom application is as impor tant as building the infrastructure itself.\n\n\n\nALL ARE WELCOME CATEGORIES:Events,Ph.D. Thesis Colloquium END:VEVENT BEGIN:VTIMEZONE TZID:Asia/Kolkata X-LIC-LOCATION:Asia/Kolkata BEGIN:STANDARD DTSTART:20250729T160000 TZOFFSETFROM:+0530 TZOFFSETTO:+0530 TZNAME:IST END:STANDARD END:VTIMEZONE END:VCALENDAR