Call for Proposals: Enterprise AI in Production, Mumbai

Saturday, 31 October 2026 · 11 AM–4 PM IST · Mumbai
Deadline to submit by Sunday, 11 October 2026, 11:59 PM IST

Venue will be announced by 7 October 2026.

Live streaming will be available for Fifthel annual members. All talks will be recorded.

Submit your proposal — https://hasgeek.com/fifthelephant/enterprise-ai-in-production-mumbai-cfp/sub

Why this meet-up

Putting AI into an enterprise means getting through security reviews, integrating with legacy systems, working within budgets, and earning the trust of the people who use it.

This meet-up brings together architects, platform engineers, data leaders, security and risk practitioners, and delivery teams to share what that work actually involves. Mumbai’s banks, insurers, payment networks, and industrial and consumer enterprises provide a setting for these conversations.

We want architectures, numbers, trade-offs, and failure stories: what you put into production, what broke, and what you changed.

Who should submit

Submit if you have worked on an enterprise AI deployment and can explain its implementation and outcomes. We welcome proposals from:

• Enterprise and solution architects designing integration patterns, shared platforms, and guardrails.
• Platform, ML, and software engineers operating LLM, retrieval, or agent workloads in production.
• Data engineering and governance practitioners making enterprise data usable, permissioned, and auditable.
• Security, risk, compliance, and audit practitioners who have approved, blocked, or reshaped deployments.
• Delivery and programme leaders who can connect implementation decisions to sustained adoption and measurable results.

We especially welcome experience from banking, financial services, insurance, payments, healthcare, telecom, manufacturing, and the public sector. First-time speakers are welcome; prior speaking experience is not a requirement.

What we want to hear about

Reference architecture and platform design

How do you build an AI platform that multiple teams can use safely? Share decisions about central versus team ownership, model gateways, routing and fallbacks, hosted versus self-hosted models, and build versus buy. Explain the boundaries you chose and how they held up.

Integration with legacy and core systems

Show how AI works with mainframes, ERPs, core banking systems, and existing APIs. Topics include safe tool access, MCP or API wrappers, transaction boundaries, idempotency, rollback, and working within change windows and release calendars.

Data readiness, retrieval, and knowledge quality

Share lessons from enterprise RAG, document permissions, freshness, metadata, deduplication, and retrieval evaluation. We also welcome structured-data and text-to-SQL case studies, including semantic layers and where they failed.

Governance, risk, and audit

Explain how you turned policy and applicable regulatory requirements into working controls. Topics include approval workflows, model inventories, audit trails, data residency, sensitive-data handling, vendor risk, and kill switches. Be specific about the requirements and organisational context of your deployment.

Evaluation, testing, and reliability

How do you know the system is working? Show evaluation sets built from real workflows, regression testing across model and data changes, tracing, incident response, and lessons from production failures.

Agents and workflow automation

We want agents with real permissions and consequences. Share how you handle identity, authorisation, human approval, prompt injection, loops, runaway costs, and silent errors—and how you measure whether automation helps.

Cost, capacity, and unit economics

Explain inference costs, GPU capacity, caching, smaller-model strategies, and cost attribution across teams. Share the production bill, the assumptions that changed, or why you decided to stop running a use case.

Adoption and operating models

Show how team structures, skills, user feedback, and operational ownership affected adoption. Connect organisational lessons to a specific implementation, including use cases that did not deliver the expected results.

Proposals may address several themes or explore one in depth.

Session formats

Choose the format that fits your proposal. The editorial team may suggest a different format during review.

What makes a strong proposal

• Production evidence: a system that is live, or was live and was retired for reasons you can explain. Include duration in production and relevant measures such as users, requests, latency, cost, or error rates.
• Architecture you can explain: components, data flows, trust boundaries, and choices you rejected.
• Honest trade-offs: what failed, how you discovered it, and what you changed.
• Clear constraints: legacy systems, security requirements, budget, headcount, procurement, or operating conditions.
• Useful takeaways: specific lessons another practitioner can apply.

We will decline sales pitches, tool tours without production context, generic frameworks, and principles without implementation evidence. Vendor practitioners are welcome when the proposal shares substantive implementation lessons and clearly discloses commercial interests.

If your work is confidential, anonymise the organisation and use ranges or relative measures where necessary. Explain these limitations in your proposal. Submit only material you are authorised to share; do not include confidential information in attachments.

What to include

1. Title and one-line pitch: what will the audience learn?
2. Abstract, 200–400 words: the problem, implementation, key decisions, and outcome.
3. Architecture summary: components, data flows, models, and tools. A diagram is welcome.
4. Production evidence: scale, time in production, measured results, and how you measured them. State where figures are approximate or anonymised.
5. Failures and trade-offs: at least one thing that went wrong or that you would do differently.
6. Three takeaways: specific things attendees can apply.
7. Audience and prerequisites: who the session is for and what they should already know.
8. Preferred format: 30-minute talk or 45-minute deep dive, including questions.
9. Speaker details: name, role, organisation, short biography, contact details, and a professional profile. Prior talks or writing are optional.
10. Disclosures and availability: commercial interests, confidentiality constraints, and your availability to speak in Mumbai on 31 October and take part in editorial preparation.

Draft slides or a short recorded walkthrough are optional.

Selection and preparation

The editorial team will review proposals for relevance, implementation depth, evidence, clarity, and usefulness to practitioners. We will also consider the balance of topics across the programme.

Shortlisted speakers may be asked for clarifications or revisions. Accepted speakers should plan for an editorial review of their slides before the meet-up.

All deadlines are in Indian Standard Time (IST).

Submit your proposal

Submit your proposal — https://hasgeek.com/fifthelephant/enterprise-ai-in-production-mumbai-cfp/sub by 11 October, 11:59 PM IST.

For questions or early feedback on an idea, contact info@hasgeek.com or 7676332020

If you have solved a difficult problem in putting AI to work inside an enterprise, we want to hear what you learned.