Data Dreams to Reality: Effective Strategies for Data Product Delivery

It is already popular opinion that more than 80% of data products dont make it into production. Those that do are asked to be exported into Excel.

What i want to discuss in this talk is a guide on how to deliver a data product from ideation to adoption. Production isnt the only goal in my opinion, it is the start. There we need to start the real work of supporting the stakeholders in effectively using the products they asked for.

We will begin with the ideation phase, how to build data wireframes, run POCs and take our stakeholders throughout the process with timely feedback loops. Then we move to development and user testing. Finally we productionalise it and periodically ask for feedback, continously improving the product so it stays relevant in the organisation.

Karma Democracy: Empowerment beyond the Ballot Box

Democracy relies on engagement, why can we not envision a democracy where voting/political engagement is tied to civic engagement on a grassroots level. The lack of participation in democracy by the younger population is well documented. The lack of participation by youth can be due to the fact that the institution of democracy is starting to feel ineffective to the Youth who prefer a more direct, loud, Substantial form of participation (Protest, Charity, Consumer Activism). What this entails is, if you tie civic engagement to political engagement then the effort people put in the democratic process extends beyond a simple vote every few years. Here people can foster grass root movements and be rewarded for it by the institution. The struggle now is to find a way to reward them. That is going to be the theme of the discussion.

The Developer's Garam Masala: Tools and Tricks for High Performance and Higher Wellbeing

In the fast-paced world of software development, productivity is key. Not to churn out code faster, but to add more value to your team, your company, and most importantly, to yourself.

In my talk, I would like to focus on these three aspects and share tips, tricks, and insights i have garnered over 7 years of working in Data Science and Tech. I realise I might not know everything that relates to a developer, that is why i enlist the help of The State of Developer Ecosystem 2023 Survey. Where over 35000 responses were submitted over 196 countries. This helped me cater my responses to what most developers according to this survey are saying about their Work, Lifestyle, and Mental Health.

Besides adding value, we need to make sure we arent overwhelming ourselves in the process. That is why in my talk I would also discuss some unusual topics about why do we need to be so productive, do we need to even code and build side projects on our time off.

This holistic approach would help developers focus on whats important to them, teams building a system that helps developers thrive, and companies by increasing the value they get out of these developers.

Cognitive Surrender: Why Games Are the Last Place We Still Think

We have quietly started to surrender our thinking. At work we let AI draft,
decide, and summarise; when we want to learn something new we accept the AI overview at the top of the search page, which can be confidently wrong more often than we admit. Each individual hand-off feels efficient. The accumulated effect is that we practise judgement less and less, and the muscle is atrophying.

This talk makes the contrarian case that games have become more important precisely because of this. A game is one of the few experiences left that
refuses to do the work for you: you cannot prompt your way past a boss, you have to read the system, form a hypothesis, fail, and adapt. You experience a
narrative instead of having it summarised. Puzzles, strategy, and difficulty are
not entertainment garnish anymore. They are deliberate, voluntary cognitive training at a time when everything else is removing friction.

I will connect three threads — cognitive surrender at work, the trap of a frictionless experience across your life, and the active problem-solving games demand — and argue that designing for productive struggle is now a feature, not a bug, in tools, teams, and life. You will leave able to spot where you have surrendered judgement, and with a concrete frame for deciding when friction is worth keeping.

Target audience: data/AI practitioners, builders, and leaders who use AI daily
and are uneasy about what it is doing to how they think. No gaming background required.
Format: 25-30 min talk; also works as a 15 min lightning.
First public delivery: new for 2026.
Technical requirements: slides + projector, audio for one short clip. No live
demo.
Takeaways (3): where you have surrendered judgement without noticing; why the frictionless life is not good for you all the time; a frame for when to keep friction on purpose.

Running Faster Nowhere: Prioritisation in the Age of AI

AI makes us feel fast. We open five threads, generate four drafts, and switch
between them and call it productivity. But much of this is task-switching taken to its limit: more motion, more half-finished work, and quietly, no prioritising.
The unspoken belief is that because we can now run faster than before, we no
longer need to choose what to run at that priorities are a constraint AI removed. They are not.
This talk separates real throughput from pseudo-productivity. Doing more things in parallel is not the same as finishing the things that matter; speed without prioritisation just produces a larger pile of incomplete output. And the bill is now coming due in two ways: the human cost of never completing anything well, and the literal one, token economics is catching up, and "spray AI at everything" is becoming visibly expensive and unsustainable.

I will argue that the right mental model is the opposite of the hype: AI lets us
move faster and better, which is exactly why prioritisation matters more, not
less. I will share a practical way to decide what deserves AI effort, how to
spend a finite token/attention budget, and how to push for complete outputs
instead of many half-ones. You will leave with a concrete filter for what to do,
what to drop, and what "done" should mean when the marginal task is nearly free to start but never free to finish.

Target audience: data/AI teams, engineers, analysts, and leads shipping with AI
day to day; anyone who feels busier and faster but not more finished.
Format: 25-30 min talk; scales to a 40 min keynote, or a 15 min lightning. First public delivery: new for 2026.
Preferred slot: main-track talk; pairs well with engineering-practice or
AI-adoption tracks.
Technical requirements: slides + projector.
Takeaways (3): how to tell throughput from task-switching; a filter for what deserves AI effort under a finite token/attention budget; redefining "done" so output is complete, not half.