Oscar Mairey

Ta-da

Training datasets for AI

2024-2025 · Head of Community

I just got off the adventure with iExec. Pretty troublesome. I'm on a ski trip with a couple of friends, thinking about the last two interviews and if I did well or not.

In the Alpes, taking some time off to reflect. Thanks Atanas for the picture!
In the Alpes, taking some time off to reflect. Thanks Atanas for the picture!

Boots on, skis clipped: time to set off. And right as we arrive home, I receive a surprise call. It's Florian, the head of operations, saying I got the role!

And more details: I'll be part of the core team. And the core team is moving to Dubai. How exciting! The move is planned around June this year.

But what job specifically?

Ta-da is a project lifted by William Simonin. In 2015 William founded Vivoka, an SDK for voice recognition capabilities. Basically so you can talk to your bluetooth speaker. That's them.

But William had an issue: the datasets they buy to train their model are both very bad and expensive. Could they use the crypto-economical incentives we see on PoS blockchains to run a system with producers of data and verifiers?

This is Ta-da. A mobile application where users are triaged to create datasets (repeat sentences, highlight images...), stake some tokens and 3 independent verifiers check the correctness of the data. If the data is bad, or the verifier lied, they lose their stake.

They raised a couple of millions of dollars, and thanks to the popularity of Owen Simonin (brother of founder William) they got a handful of users pretty rapidly.

Already at the end of 2023 on public beta, users were flowing in to complete tasks. MultiversX is the blockchain on which the application ran
Already at the end of 2023 on public beta, users were flowing in to complete tasks. MultiversX is the blockchain on which the application ran

With my past experience in marketing and community, I was set to work with Pierre (we call him Pepe) on the community (hear: user base) of the product.

Turbulence ahead!

Ta-da shares the same offices as all Simonin's brothers companies. So naturally I met Owen early on in the kitchen when the team was drinking coffee. Needless to say I got star struck. Moving on.

The team had just released its token, and we were growing rapidly. But the team (meaning Pepe and I) had no real infrastructure to work off of. So I installed a couple of easy wins: project management, to-do lists, social planners, weekly community meetups, ambassador programs...

Couple of months by, the social accounts doubled in size, community retention was very high with 100 members joining weekly meetups, and we expanded the user base to several new countries (Turkey, Italy, Romania...) with specialized community managers.

The announcing video of the Turkish launch. A great success!
The announcing video of the Turkish launch. A great success!

The informal pact with Pepe was I knew what I was doing professionally, but inter-personally I was still rough around the edges. He'd cover for the mistakes I'd make with colleagues, and the team would get the wins.

And I did make some mistakes. Glad he was there to protect my ass.

Come to Dubai, Habibi!

By May, the first batch of employees flew out to Dubai. I was not part of it. I fully trusted the team to be part of the second (and only) batch.

The real test was the public launch of the app, from public beta to actual money flowing in. And it was a clear success.

The post announcing the end of the public beta. Notice great engagement and views!
The post announcing the end of the public beta. Notice great engagement and views!

Second batch came, and I had my flight tickets!

The first off-site planned one week after we arrived with the team. Had to resist the urge to catch the falcon.
The first off-site planned one week after we arrived with the team. Had to resist the urge to catch the falcon.

So what is he doing?

I was in the habit to work late nights during the first few weeks. Anyway, there was nothing else to do as I did not know the city nor had friends here.

It was also the opportunity to grab quick five-minutes chats here and there with William. And one night, as I was still at work, he waved goodbye and chatted a bit. It went something like:

W: "What are you working on?"
O: "I'm working on next week's scheduled posts"
W: "And Pepe is working for the AMA tomorrow?"
O: "Nope, I already took care of that. We have the invite list and subjects."
W: "So what is Pepe doing?"
O: "He's managing me!"

At the end of the week, I was promoted to head of our four-person team. It doubled my salary.

Working with Florian

After some months working in this position, it hit me I've learnt everything there is to know about community management and marketing. At least in this role.

So I've had a conversation with the C-levels to join the sales team. To learn more about sales as it is a great skill to have. Florian, head of sales, said sure.

For a solid 3 months, I had serious leads for our datasets with a major EV manufacturer among others lesser-known startups. Whereas the period was very dry for the whole team, I did manage to sell a couple of our services to some companies.

Turbulences Again

Around October, things started to get flaky. We could feel the tension around the office as the company needed to shrink back to focus on its core business: producing datasets, and selling them.

One night, Florian comes into the office and tells me a colleague will not continue with us. But not to worry, because they will find a place for me. I thanked Florian, but told him not to make great efforts to keep me in the company.

By then I had gradually started working with Gregory, whom I'd met a few months prior... And one of his companies Chainraizer just raised a great amount of money.