Suggestions
Dom Pitot 🇿🇦 🚀 🇺🇲
The Texan Saffa | Founder of Joyforge Labs | Partner at Axis Data
Dominique de la Beaujardiere is a highly accomplished professional with a diverse background in finance, engineering, and technology. An alumnus of the prestigious University of Oxford, she earned both her Msc in Economic and Social History, with a focus on Economic History, and her Msc in Financial Economics, with a focus on Corporate Finance, Econometrics, Asset Pricing, Micro and Macro Economics. Prior to this, Dominique completed a Bachelor's degree in Electrical and Computer Engineering from the University of Cape Town.
Dominique's career has spanned across various reputable organizations, where she held key executive and managerial roles in Africa and globally. She is currently the Chief Executive Officer at Inoxico Pty Ltd, a cybersecurity company that specializes in due diligence, AML investigations, and third-party risk services among other things. Dominique's leadership experience at Inoxico stems from her previous roles as the Executive Head of Inoxico Core and Chief Operating Officer. Before joining Inoxico, she served as an Account Executive at ITC Global, where she worked with clients in hazardous work environments, providing them with cutting-edge satellite communication solutions.
She is also the former Founder and Managing Director of Mining Atlas, a digital repository of mining and exploration projects worldwide. During her tenure, Dominique was responsible for the overall strategy and operations of the company. She made significant contributions, including expanding product offerings, increasing global coverage, and securing institutional partnerships.
Before these positions, Dominique was a Regional Sales Manager (Africa) at Siemens Industry and a Global Commodity Manager at Anglo American. She began her career in the mining sector as a Process Control Engineer at Anglo American as part of a Graduate Development Program. Dominique's extensive experience in various sectors has equipped her with versatile skills, including supply chain management, corporate finance analysis, and transformation management.
In her spare time, Dominique volunteers as a Youth Program Manager for Habitat for Humanity International, where she contributes to improving and building communities worldwide. She is also a former Transformation Manager at the University of Cape Town Student Council.
Highlights
Drive or Fly?
These days I often need to get to Dallas from Houston Its a 5 hour drive or a 1 hour flight.
But flying has a lot of complications (airports, luggage, security, delays) and actually takes 4 hours door-to-door.
Which do you prefer?
I had a similar regular trip in South Africa: Johannesburg to Pietermaritzburg. And I drove it every time, despite that it's tiring and more dangerous.
I can't wait till we have self-driving cars!
Getting Ai from Prototype to Production Is nearly impossible Let's be honest 😯 We have failed to convert our Ai experiments into reliable solutions There are three main reasons for this. 1️⃣ Non-Determinism Most Ai models (especially LLMs) give different outputs for the same inputs. Just ask ChatGPT to tell you a joke and every time you get a different (silly) joke. That’s fun for a demo 🎉 🕺 But relying on an Ai model to calculate a key metric? or execute a key task? You want consistent outcomes. You want it to be deterministic. 2️⃣ Non-Explainability Most Ai models cannot tell you how it arrived at its answer. The explanation would be stochastic gobbledygook. That’s fine when the answer is what you expect. But when the answer is a surprise 😨 you need to know why ⁉ Have your clients ever asked why your product gave a crazy answer? If you cannot tell them why, they lose trust. 3️⃣ Production Readiness Once you put an Ai prototype into a real "production", it breaks down quickly. The model itself performs fine but data pipelines and automations are brittle and break as soon as edge-cases and noise are fed in.
Sustaining the entire ecosystem is incredibly complex. Bits of python feed bits of R, fed by ETLs from different databases across multiple domains When something goes wrong, its takes forever to unravel the web. These three factors erode confidence until the shiny model is thrown into the dumpster. Game over 🎮 But going from Prototype to Production is NOT impossible. My team at Trade Shield achieved this feat 💪 I speak from experience having faced disaster many times while delivering predictive results that saved millions 💰 And today I partner with Palantir Technologies who have developed an incredible data orchestration platform that makes Ai models in production a reality. When I found Foundry, it was "Love-at-First-Build!" 😍
DM me if you want to find out more about Palantir Technologies and how we support clients to use Foundry for data orchestration.
