The Fifth Elephant 2026 Winter Edition

Call for Data Problems and Proposals

4 & 5 December 2026
Working with data: what changes, what endures, and what becomes possible.

A conference for people who use data to understand the world, investigate questions and make decisions—in industry, research, journalism, public policy and the social sector.

Submit your proposal by 11 October 2026.

Register to attend

Bring a problem you have been thinking about

Two teams report different numbers for the same metric. A clean-looking public dataset leads to the wrong conclusion. An experiment passes every test, until you discover what the tests missed. A supplier delay raises a question: what breaks next?

The Fifth Elephant Winter Edition is about the work behind answering such questions: understanding how data was generated, deciding what to measure, choosing methods, testing explanations and knowing when a result deserves another look.

AI is changing what we can attempt and how quickly we can explore it. We want to hear what this makes possible in your work, where it falls short, and which principles matter as much as ever. Your proposal can involve statistics, spreadsheets, visualization, machine learning, LLMs or a combination. Using AI is not a requirement.

Bring new ideas, familiar problems approached differently, surprising findings, failed attempts and work in progress. You do not need a finished success story.

Who can propose to speak/showcase/teach

We welcome analysts, data scientists, statisticians, researchers, journalists, visualization practitioners, and people who work with data in business, government and non-profits.

Your domain might be retail, finance, education, healthcare, economics, elections, manufacturing, the arts or something we have not listed. What matters is that you can make the problem understandable and share something useful beyond your own setting.

The Winter Edition is also a place to find collaborators: people with a different dataset, a useful method, domain knowledge, or a question that changes how you see your work.

What we are looking for

We are interested in proposals that explore questions such as:

• Can we trust this data? How collection, sampling, annotation, missing information or changing definitions affected your conclusions—and how you discovered it.
• Are we asking and measuring the right things? How you chose a question, designed an experiment, evaluated a model or found that a metric was rewarding the wrong outcome.
• How did you investigate something unexpected? Show the observation that puzzled you, the explanations you tested, and what you learned or still cannot explain.
• What becomes possible with newer methods? Show an analysis, experiment or application that was previously difficult, expensive or impractical, and how you checked the results.
• How does analysis become a decision? Explore business constraints, domain knowledge, competing interpretations and the organizational realities that shape how findings are used.
• What can others do with your work? Share a dataset, visualization, research project or open problem that could prompt further investigation or collaboration.

The focus is data practice and analytical reasoning. Infrastructure, pipelines, deployment and tool implementation are outside this edition’s scope when they are the main subject. Technical detail is welcome when it helps explain the investigation, evidence or lesson.

Help the audience learn from your process

Start with the problem and the data. Explain what you tried, why you made those choices, what happened, and where the limitations remain. Make room for the false starts and judgments that a polished result can hide.

A useful question to answer in your proposal is: what will someone in the audience be able to do or think through differently after your session?

Show enough of the work for people to examine your reasoning. If you cannot share the original data, explain what you can show through anonymized examples or clearly labelled synthetic data, and what those examples cannot establish.

Unfinished work is welcome. Be clear about what you have already tried, what evidence you have, and what remains a hypothesis or an open question.

Ways to participate

Practitioner talks · 25 minutes + 5 mins Q&A

Explore a problem, method or investigation in depth. Use concrete examples to explain your reasoning, findings and limitations, and help people recognize where the lessons apply in their own work.

You can present completed work, a failed approach, or an investigation still in progress.

Showcases · 20 minutes

Show one or two interesting things you have done with data: an analysis, visualization, experiment or practical application.

Your project need not be in production or introduce a new technique. Give us a focused look at what you made or discovered, how you approached it, and why it may inspire someone else.

Problem wall

Bring a problem you would like help with - or a dataset looking for worthwhile questions. Explain what you know, what you have tried, where you are stuck, and which ideas, expertise or collaborators would help.

Birds of a Feather (BOF) sessions

Propose a discussion that benefits from people comparing experiences and working through questions together. Tell us who should join, what you want to explore, and how you will get the room participating.

Posters

Share research, work in progress, charts, maps or a project visually. Give people something they can examine, ask questions about and discuss with you between sessions.

The first confirmed talk for Winter edition

When a fabric mill is late, what breaks next? — Abhijith Neekaje (Beyond Vectors)
A delayed fabric order can affect products, stores and planned promotions. Abhijith’s retail case study examines how to represent those dependencies in data and check whether proposed responses meet business constraints. It uses knowledge graphs and neurosymbolic AI to explore a broader question: what must we make explicit before an AI recommendation can be verified?

Not sure where your idea fits?

Submit it anyway. You can propose more than one idea, and you do not need to settle on a format before starting the conversation. The editors can help you refine the focus and identify a suitable format.

The final scheduled sounding board call is Friday, 9 October, 12:00–12:40 PM IST. Bring an idea or a question and discuss it with the editors.

Join the sounding board call
Submit a proposal

Meet the editors

• Karthik Shashidhar — AI and data consultant
• S Anand — LLM Psychologist at Straive
• Prasanna Bhogale — Founder at Romulan AI; author

Deadline

Submit by 11 October 2026. Earlier submissions give us more time for discussion and feedback.

The planned programme includes six practitioner talks and six showcases, alongside the participatory formats above.

Questions?

Email info@hasgeek.com or call/text 7676332020