StarZero All shows Try the podcast clipper →

Sep 17, 2018 · 25m · mad

The Launch of Dataiku 5 // Florian Douetteau, Dataiku (FirstMark's Data Driven NYC)

Florian Douetteau · 18m spoken Matt Turck · 57s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

At FirstMark's Data Driven NYC, Dataiku Founder and CEO Florian Douetteau outlines how modern enterprises can scale AI through an inclusive, people-driven approach and introduces the architecture, product credos, and features of Dataiku 5.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 4.4% of the talking time here. How this is scored →

Matt as informed peer 0.4 Guest teaching 4.9 Guest disagreement 1.3 Matt pushing back 0.3
05100:0010:0020:000:00–2:09 · Matt as informed peer 0/10 Event Title and Speaker Introductions Florian opens the meetup monologue by playfully suggesting renaming 'Data Driven NYC' to 'People Driven NYC' because human execution matters more than data availability. Host does not speak during this presentation segment.2:09–4:47 · Matt as informed peer 0/10 Scaling Data Science Initiatives Across Global Teams Florian narrative-builds around the archetype of 'Gary' growing a data team from a single laptop project to a global team. He contrasts the 'elite PhD' software approach with the collaborative business-plus-tech model.4:47–7:18 · Matt as informed peer 0/10 Integrated Platforms vs. Point Solutions ('Mutts Are the Best') Florian rejects the popular 'best of breed' point solution software approach, arguing through a dog analogy that 'mutts are the best' integrated platforms. Host remains off-stage during the presentation.7:18–9:29 · Matt as informed peer 0/10 Cross-Industry Customer Adoption and Dataiku 5 Launch Florian introduces Dataiku 5 and its first product credo regarding separating specification from execution. He uses dry humor to poke fun at software 'hype' driving adoption of Docker and Kubernetes.9:29–11:52 · Matt as informed peer 0/10 Dataiku Credo II — Automated Machine Learning and Gradients of Complexity Florian presents Credo II on automated machine learning, comparing it to a camera with gradients between automatic mode and full manual expert control. Host does not participate.11:52–15:00 · Matt as informed peer 0/10 Agility, Open Source Integration, and Centralized Metadata Hub Florian presents Credo III on metadata centralization, referencing open source tools like Airflow and JupyterLab while noting their limitations, and finishes with company takeaway offers.15:00–25:20 · Matt as informed peer 3/10 Fireside Q&A Session with Matt Turck Matt Turck hosts a Q&A session asking structured questions on enterprise adoption lifecycles and industry verticals, and clarifies a vague audience question on project risks. Florian delivers authoritative data science management answers.0:00–2:09 · Guest teaching 3/10 Event Title and Speaker Introductions Florian opens the meetup monologue by playfully suggesting renaming 'Data Driven NYC' to 'People Driven NYC' because human execution matters more than data availability. Host does not speak during this presentation segment.2:09–4:47 · Guest teaching 5/10 Scaling Data Science Initiatives Across Global Teams Florian narrative-builds around the archetype of 'Gary' growing a data team from a single laptop project to a global team. He contrasts the 'elite PhD' software approach with the collaborative business-plus-tech model.4:47–7:18 · Guest teaching 5/10 Integrated Platforms vs. Point Solutions ('Mutts Are the Best') Florian rejects the popular 'best of breed' point solution software approach, arguing through a dog analogy that 'mutts are the best' integrated platforms. Host remains off-stage during the presentation.7:18–9:29 · Guest teaching 5/10 Cross-Industry Customer Adoption and Dataiku 5 Launch Florian introduces Dataiku 5 and its first product credo regarding separating specification from execution. He uses dry humor to poke fun at software 'hype' driving adoption of Docker and Kubernetes.9:29–11:52 · Guest teaching 5/10 Dataiku Credo II — Automated Machine Learning and Gradients of Complexity Florian presents Credo II on automated machine learning, comparing it to a camera with gradients between automatic mode and full manual expert control. Host does not participate.11:52–15:00 · Guest teaching 5/10 Agility, Open Source Integration, and Centralized Metadata Hub Florian presents Credo III on metadata centralization, referencing open source tools like Airflow and JupyterLab while noting their limitations, and finishes with company takeaway offers.15:00–25:20 · Guest teaching 6/10 Fireside Q&A Session with Matt Turck Matt Turck hosts a Q&A session asking structured questions on enterprise adoption lifecycles and industry verticals, and clarifies a vague audience question on project risks. Florian delivers authoritative data science management answers.0:00–2:09 · Guest disagreement 1/10 Event Title and Speaker Introductions Florian opens the meetup monologue by playfully suggesting renaming 'Data Driven NYC' to 'People Driven NYC' because human execution matters more than data availability. Host does not speak during this presentation segment.2:09–4:47 · Guest disagreement 1/10 Scaling Data Science Initiatives Across Global Teams Florian narrative-builds around the archetype of 'Gary' growing a data team from a single laptop project to a global team. He contrasts the 'elite PhD' software approach with the collaborative business-plus-tech model.4:47–7:18 · Guest disagreement 2/10 Integrated Platforms vs. Point Solutions ('Mutts Are the Best') Florian rejects the popular 'best of breed' point solution software approach, arguing through a dog analogy that 'mutts are the best' integrated platforms. Host remains off-stage during the presentation.7:18–9:29 · Guest disagreement 2/10 Cross-Industry Customer Adoption and Dataiku 5 Launch Florian introduces Dataiku 5 and its first product credo regarding separating specification from execution. He uses dry humor to poke fun at software 'hype' driving adoption of Docker and Kubernetes.9:29–11:52 · Guest disagreement 1/10 Dataiku Credo II — Automated Machine Learning and Gradients of Complexity Florian presents Credo II on automated machine learning, comparing it to a camera with gradients between automatic mode and full manual expert control. Host does not participate.11:52–15:00 · Guest disagreement 1/10 Agility, Open Source Integration, and Centralized Metadata Hub Florian presents Credo III on metadata centralization, referencing open source tools like Airflow and JupyterLab while noting their limitations, and finishes with company takeaway offers.15:00–25:20 · Guest disagreement 1/10 Fireside Q&A Session with Matt Turck Matt Turck hosts a Q&A session asking structured questions on enterprise adoption lifecycles and industry verticals, and clarifies a vague audience question on project risks. Florian delivers authoritative data science management answers.0:00–2:09 · Matt pushing back 0/10 Event Title and Speaker Introductions Florian opens the meetup monologue by playfully suggesting renaming 'Data Driven NYC' to 'People Driven NYC' because human execution matters more than data availability. Host does not speak during this presentation segment.2:09–4:47 · Matt pushing back 0/10 Scaling Data Science Initiatives Across Global Teams Florian narrative-builds around the archetype of 'Gary' growing a data team from a single laptop project to a global team. He contrasts the 'elite PhD' software approach with the collaborative business-plus-tech model.4:47–7:18 · Matt pushing back 0/10 Integrated Platforms vs. Point Solutions ('Mutts Are the Best') Florian rejects the popular 'best of breed' point solution software approach, arguing through a dog analogy that 'mutts are the best' integrated platforms. Host remains off-stage during the presentation.7:18–9:29 · Matt pushing back 0/10 Cross-Industry Customer Adoption and Dataiku 5 Launch Florian introduces Dataiku 5 and its first product credo regarding separating specification from execution. He uses dry humor to poke fun at software 'hype' driving adoption of Docker and Kubernetes.9:29–11:52 · Matt pushing back 0/10 Dataiku Credo II — Automated Machine Learning and Gradients of Complexity Florian presents Credo II on automated machine learning, comparing it to a camera with gradients between automatic mode and full manual expert control. Host does not participate.11:52–15:00 · Matt pushing back 0/10 Agility, Open Source Integration, and Centralized Metadata Hub Florian presents Credo III on metadata centralization, referencing open source tools like Airflow and JupyterLab while noting their limitations, and finishes with company takeaway offers.15:00–25:20 · Matt pushing back 2/10 Fireside Q&A Session with Matt Turck Matt Turck hosts a Q&A session asking structured questions on enterprise adoption lifecycles and industry verticals, and clarifies a vague audience question on project risks. Florian delivers authoritative data science management answers.

speaking balance: gold is Matt, purple is the guest (3 minute bins)

0:00 · Matt 0% · guest 100%0:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%15:00 · Matt 23.3% · guest 76.7%15:00 · Matt 23.3% · guest 76.7%18:00 · Matt 7.8% · guest 92.2%18:00 · Matt 7.8% · guest 92.2%21:00 · Matt 3.6% · guest 96.4%21:00 · Matt 3.6% · guest 96.4%24:00 · Matt 9.5% · guest 90.5%24:00 · Matt 9.5% · guest 90.5%
Sharpest disagreement ▶ 5:15 Rejecting 'Best of Breed' Point Solutions

Florian contrarianly dismisses the popular industry paradigm of 'best of breed' software point solutions, using a dog analogy to argue that integrated platforms ('mutts') are vastly superior.

 Hardest push from Matt ▶ 18:52 Reframing Audience Question on Project Failures

Matt Turck steps in when an audience question is too broad, refocusing the topic specifically on enterprise project failure modes and execution risks.

 Biggest teaching moment ▶ 9:55 Graduated Control in Automated Machine Learning

Florian educates the audience on AutoML by reframing it away from a binary black box into a spectrum of manual and automatic modes using a camera control analogy.

 Matt holds his own ▶ 16:02 Probing Macro Enterprise Adoption Curves

Matt Turck demonstrates domain fluency by posing a well-structured macroeconomic question about adoption curve positioning across enterprise verticals.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
Event Title and Speaker Introductions 0310 Florian opens the meetup monologue by playfully suggesting renaming 'Data Driven NYC' to 'People Driven NYC' because human execution matters more than data availability. Host does not speak during this presentation segment.
Scaling Data Science Initiatives Across Global Teams 0510 Florian narrative-builds around the archetype of 'Gary' growing a data team from a single laptop project to a global team. He contrasts the 'elite PhD' software approach with the collaborative business-plus-tech model.
Integrated Platforms vs. Point Solutions ('Mutts Are the Best') 0520 Florian rejects the popular 'best of breed' point solution software approach, arguing through a dog analogy that 'mutts are the best' integrated platforms. Host remains off-stage during the presentation.
Cross-Industry Customer Adoption and Dataiku 5 Launch 0520 Florian introduces Dataiku 5 and its first product credo regarding separating specification from execution. He uses dry humor to poke fun at software 'hype' driving adoption of Docker and Kubernetes.
Dataiku Credo II — Automated Machine Learning and Gradients of Complexity 0510 Florian presents Credo II on automated machine learning, comparing it to a camera with gradients between automatic mode and full manual expert control. Host does not participate.
Agility, Open Source Integration, and Centralized Metadata Hub 0510 Florian presents Credo III on metadata centralization, referencing open source tools like Airflow and JupyterLab while noting their limitations, and finishes with company takeaway offers.
Fireside Q&A Session with Matt Turck 3612 Matt Turck hosts a Q&A session asking structured questions on enterprise adoption lifecycles and industry verticals, and clarifies a vague audience question on project risks. Florian delivers authoritative data science management answers.

Statements from this episode (11)

Opinion
Douetteau: People, not data, are the real challenge in tech today
“Data, it's here. It's no longer a problem. And I think the real challenge today is about, like, and the change in data, like, the people will make it happen.”
Florian Douetteau Sep 17, 2018 ▶ 0:30
Assertion Not checkable as stated
Douetteau: Every recent data science effort started with an individual champion
“Every data science story I heard in the last few years actually started with someone.”
Florian Douetteau Sep 17, 2018 ▶ 1:12
Insight
Douetteau: Enterprise data science fails without combining business insights with technical expertise
“Data science is actually a mix of business and tech, and the challenge is actually to make business people and tech people get on board in data science together and work collaboratively on project, because you cannot really deliver data science without having …”
Florian Douetteau Sep 17, 2018 ▶ 4:22
Opinion
Douetteau: Best-of-breed enterprise software point solutions are overrated
“I think best of breed is, like, so overrated.”
Florian Douetteau Sep 17, 2018 ▶ 5:38
Insight
Douetteau: Founders must implement strong product beliefs like maniacs
“When you want to build software, well, you need to listen to your customers, but you also need to have strong beliefs and try to implement them in your software like a maniac.”
Florian Douetteau Sep 17, 2018 ▶ 7:56
Insight
Douetteau: Data platforms must decouple specification from execution
“First part of our credo is that, well, you should kind of separate specification and execution as much as possible in data.”
Florian Douetteau Sep 17, 2018 ▶ 8:13
Prediction Not checkable as stated
Douetteau: Machine learning will be automated in the next few years
“Credo number three, two, actually, is that I think that, well, machine learning is to be automated in the next few years.”
Florian Douetteau Sep 17, 2018 ▶ 9:30
Opinion
Douetteau: Airflow and JupyterLab don't go far enough for enterprise data science
“Those two projects are, like, very interesting, but to some extent don't go far enough, because what you really want, at the end of the day, is the ability to capture, within the platform, everything related to, ah, to your data science project.”
Florian Douetteau Sep 17, 2018 ▶ 12:21
Prediction Not checkable as stated
Douetteau: Global manufacturing will adopt data science within ten years
“So I guess that it will also get to global manufacturing in the next 10 years, probably.”
Florian Douetteau Sep 17, 2018 ▶ 17:40
Insight
Douetteau: Most enterprise data projects fail due to missing or poor data
“The main risk of a data-driven world is actually not being data-driven because you don't have the data. I mean, most of the project I saw failing dramatically where because of data not being available or not being in the format or the quality that was expected…”
Florian Douetteau Sep 17, 2018 ▶ 19:09
Insight
Douetteau: Companies cannot hire a data scientist without already having one
“You can't hire a data scientist without a data scientist, meaning you don't even know what the hiring process is.”
Florian Douetteau Sep 17, 2018 ▶ 24:44
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more