BEGIN:VCALENDAR VERSION:2.0 PRODID:-//HasGeek//NONSGML Funnel//EN DESCRIPTION:Share what it takes to run AI inside an enterprise: architectu res\, trade-offs\, and lessons from production. X-WR-CALDESC:Share what it takes to run AI inside an enterprise: architect ures\, trade-offs\, and lessons from production. NAME:Enterprise AI in Production: Mumbai Call for Proposals X-WR-CALNAME:Enterprise AI in Production: Mumbai Call for Proposals REFRESH-INTERVAL;VALUE=DURATION:PT12H SUMMARY:Enterprise AI in Production: Mumbai Call for Proposals TIMEZONE-ID:Asia/Kolkata X-PUBLISHED-TTL:PT12H X-WR-TIMEZONE:Asia/Kolkata BEGIN:VEVENT SUMMARY:Enterprise AI in Production: Mumbai Call for Proposals DTSTART:20261031T053000Z DTEND:20261031T103000Z DTSTAMP:20261023T034918Z UID:session/BcvEBiutsJvDkuem1dPS9t@hasgeek.com SEQUENCE:13 CREATED:20261002T033044Z DESCRIPTION:# Call for Proposals: Enterprise AI in Production\, Mumbai\n\n **Saturday\, 31 October 2026 · 11 AM–4 PM IST · Mumbai** \n**Deadline to submit by Sunday\, 18 October 2026\, 11:59 PM IST**\n\nVenue will be a nnounced shortly.\n\nLive streaming will be available for [Fifthel annual members](https://hasgeek.com/fifthelephant#memberships). All talks will be recorded.\n\n**Submit your proposal — https://hasgeek.com/fifthelephant /enterprise-ai-in-production-mumbai-cfp/sub**\n\n## Why this meet-up\nPutt ing AI into an enterprise means getting through security reviews\, integra ting with legacy systems\, working within budgets\, and earning the trust of the people who use it.\n\nThis meet-up brings together architects\, pla tform engineers\, data leaders\, security and risk practitioners\, and del ivery teams to share what that work actually involves. Mumbai’s banks\, insurers\, payment networks\, and industrial and consumer enterprises prov ide a setting for these conversations.\n\nWe want architectures\, numbers\ , trade-offs\, and failure stories: what you put into production\, what br oke\, and what you changed.\n\n## Who should submit\nSubmit if you have wo rked on an enterprise AI deployment and can explain its implementation and outcomes. We welcome proposals from:\n\n- Enterprise and solution archite cts designing integration patterns\, shared platforms\, and guardrails.\n- Platform\, ML\, and software engineers operating LLM\, retrieval\, or age nt workloads in production.\n- Data engineering and governance practitione rs making enterprise data usable\, permissioned\, and auditable.\n- Securi ty\, risk\, compliance\, and audit practitioners who have approved\, block ed\, or reshaped deployments.\n- Delivery and programme leaders who can co nnect implementation decisions to sustained adoption and measurable result s.\n\nWe especially welcome experience from banking\, financial services\, insurance\, payments\, healthcare\, telecom\, manufacturing\, and the pub lic sector. First-time speakers are welcome\; prior speaking experience is not a requirement.\n\n## What we want to hear about\n\n### Reference arch itecture and platform design\nHow do you build an AI platform that multipl e teams can use safely? Share decisions about central versus team ownershi p\, model gateways\, routing and fallbacks\, hosted versus self-hosted mod els\, and build versus buy. Explain the boundaries you chose and how they held up.\n\n### Integration with legacy and core systems\nShow how AI work s with mainframes\, ERPs\, core banking systems\, and existing APIs. Topic s include safe tool access\, MCP or API wrappers\, transaction boundaries\ , idempotency\, rollback\, and working within change windows and release c alendars.\n\n### Data readiness\, retrieval\, and knowledge quality\nShare lessons from enterprise RAG\, document permissions\, freshness\, metadata \, deduplication\, and retrieval evaluation. We also welcome structured-da ta and text-to-SQL case studies\, including semantic layers and where they failed.\n\n### Governance\, risk\, and audit\nExplain how you turned poli cy and applicable regulatory requirements into working controls. Topics in clude approval workflows\, model inventories\, audit trails\, data residen cy\, sensitive-data handling\, vendor risk\, and kill switches. Be specifi c about the requirements and organisational context of your deployment.\n\ n### Evaluation\, testing\, and reliability\nHow do you know the system is working? Show evaluation sets built from real workflows\, regression test ing across model and data changes\, tracing\, incident response\, and less ons from production failures.\n\n### Agents and workflow automation\nWe wa nt agents with real permissions and consequences. Share how you handle ide ntity\, authorisation\, human approval\, prompt injection\, loops\, runawa y costs\, and silent errors—and how you measure whether automation helps .\n\n### Cost\, capacity\, and unit economics\nExplain inference costs\, G PU capacity\, caching\, smaller-model strategies\, and cost attribution ac ross teams. Share the production bill\, the assumptions that changed\, or why you decided to stop running a use case.\n\n### Adoption and operating models\nShow how team structures\, skills\, user feedback\, and operationa l ownership affected adoption. Connect organisational lessons to a specifi c implementation\, including use cases that did not deliver the expected r esults.\n\nProposals may address several themes or explore one in depth.\n \n## Session formats\n\n| Format | Total session time\, including question s | Best suited to |\n| --- | --- | --- |\n| Talk | 30 minutes | One syste m or decision\, with architecture\, trade-offs\, and outcomes |\n| Deep di ve | 45 minutes | A complex implementation with diagrams\, detailed exampl es\, or a walkthrough |\n\nChoose the format that fits your proposal. The editorial team may suggest a different format during review.\n\n## What ma kes a strong proposal\n\n- **Production evidence:** a system that is live\ , or was live and was retired for reasons you can explain. Include duratio n in production and relevant measures such as users\, requests\, latency\, cost\, or error rates.\n- **Architecture you can explain:** components\, data flows\, trust boundaries\, and choices you rejected.\n- **Honest trad e-offs:** what failed\, how you discovered it\, and what you changed.\n- * *Clear constraints:** legacy systems\, security requirements\, budget\, he adcount\, procurement\, or operating conditions.\n- **Useful takeaways:** specific lessons another practitioner can apply.\n\nWe will decline sales pitches\, tool tours without production context\, generic frameworks\, and principles without implementation evidence. Vendor practitioners are welc ome when the proposal shares substantive implementation lessons and clearl y discloses commercial interests.\n\nIf your work is confidential\, anonym ise the organisation and use ranges or relative measures where necessary. Explain these limitations in your proposal. Submit only material you are a uthorised to share\; do not include confidential information in attachment s.\n\n## What to include\n1. **Title and one-line pitch:** what will the a udience learn?\n2. **Abstract\, 200–400 words:** the problem\, implement ation\, key decisions\, and outcome.\n3. **Architecture summary:** compone nts\, data flows\, models\, and tools. A diagram is welcome.\n4. **Product ion evidence:** scale\, time in production\, measured results\, and how yo u measured them. State where figures are approximate or anonymised.\n5. ** Failures and trade-offs:** at least one thing that went wrong or that you would do differently.\n6. **Three takeaways:** specific things attendees c an apply.\n7. **Audience and prerequisites:** who the session is for and w hat they should already know.\n8. **Preferred format:** 30-minute talk or 45-minute deep dive\, including questions.\n9. **Speaker details:** name\, role\, organisation\, short biography\, contact details\, and a professio nal profile. Prior talks or writing are optional.\n10. **Disclosures and a vailability:** commercial interests\, confidentiality constraints\, and yo ur availability to speak in Mumbai on 31 October and take part in editoria l preparation.\n\nDraft slides or a short recorded walkthrough are optiona l. \n\n## Selection and preparation\nThe editorial team will review propos als for relevance\, implementation depth\, evidence\, clarity\, and useful ness to practitioners. We will also consider the balance of topics across the programme.\n\nShortlisted speakers may be asked for clarifications or revisions. Accepted speakers should plan for an editorial review of their slides before the meet-up.\n\n| Milestone | Date |\n| --- | --- |\n| Venue announced | By Wednesday\, 7 October |\n| Proposals close | Sunday\, 11 O ctober\, 11:59 PM IST |\n| Clarifications and proposal revisions | 12–14 October |\n| Selection decisions sent | Thursday\, 15 October |\n| Select ed speakers confirm participation | Friday\, 16 October |\n| Schedule anno unced | Monday\, 19 October |\n| First slide drafts due | Friday\, 23 Octo ber |\n| Editorial reviews and walkthroughs | 24–28 October |\n| Final s lides due | Thursday\, 29 October |\n| Meet-up in Mumbai | Saturday\, 31 O ctober 2026\, 11 AM–4 PM IST |\n\nAll deadlines are in Indian Standard T ime (IST).\n\n## Submit your proposal\n\n**Submit your proposal — https: //hasgeek.com/fifthelephant/enterprise-ai-in-production-mumbai-cfp/sub** b y **18 October\, 11:59 PM IST**.\n\nFor questions or early feedback on an idea\, contact info@hasgeek.com or 7676332020\n\nIf you have solved a diff icult problem in putting AI to work inside an enterprise\, we want to hear what you learned. LAST-MODIFIED:20261014T123227Z LOCATION:Mumbai - https://hasgeek.com/fifthelephant/enterprise-ai-in-produ ction-mumbai-cfp/ ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/enterprise-ai-in-production-mumbai-c fp/ BEGIN:VALARM ACTION:display DESCRIPTION:Enterprise AI in Production: Mumbai Call for Proposals in 5 mi nutes TRIGGER:-PT5M END:VALARM END:VEVENT END:VCALENDAR