Peroptyx enhances machine learning training data and model performance
Please introduce yourself and your startup Peroptyx to our readers!
Peroptyx is a technology company that enhances machine learning training data and model performance for consumer-facing internet platforms and services.
Our team has solved global data quality and model evaluation problems for the major internet platforms and search engines over the past 15 years. Our platform addresses the global sourcing, retention, data privacy, security and personal identity issues that exist with current ‘human-in-the-loop’ solutions for training data and model evaluation.
Peroptyx is the unique combination of both that will help businesses accelerate their global Machine Learning deployments in the 2020s.
Prior to co-founding Peroptyx, I was the founding GM and VP of Lionbridge AI, building that business from startup to reach $100M in revenues with a remote working team spanning over 100 countries. Having begun my career as a software engineer with Microsoft, I could see a significant opportunity to accelerate growth for a new generation of born global businesses who were going ‘straight to Machine Learning’ to generate exponential value from their data.
How did you get the idea of Peroptyx?
The idea for the name came while watching the almost magical, synchronous flashing of swarms of Pteroptyx (sic) fireflies as they congregated around mangrove trees, mostly in southeast Asia! It is an unforgettable experience that also happens to deliver significant economic benefits to the region. This visual spectacle became synonymous with our passion for operational synchronicity and swarm intelligence. The idea for the company was born of the belief that my co-founders Maeve Bleahene, Internet pioneer Dr. Dennis Jennings and I shared regarding how the everyday human experience was not being adequately reflected by consumer-facing AI systems, or in the Machine Learning models and data underpinning them.
Our combined experience over the prior decade showed conclusively that the quality of the AI experience will ultimately be determined by the quality of the human contribution to the data powering AI. Together, we recognised the huge potential in Machine Learning for companies prepared to make the leap, to put data at the heart of their approach to developing market-facing platforms, applications and services.
Why did you decide to start Peroptyx?
As mentioned previously, I had successfully scaled a global business within a corporation as an ‘intrapreneur’. Throughout that journey, the joy of working with and learning from industry leaders and global technology brands was highly motivating – we experimented and collaborated to solve some challenging problems. Over time, It became clear to me that a significant number of next-generation enterprises wanted to or were preparing to embark on the journey to become more data-driven, but were missing key guidance and insights around operationalising machine learning deployments at a global scale. At the same time, our vision around the future of data was not shared by a corporation that had become overburdened with a bricks-and-mortar-based business model.
With that, we decided to start a company fully focused on delivering sustainable value to enterprises taking a data-centric, cloud-first approach to building and deploying ML in their online products and services.
We gathered a team and together we went to work on building the company, born in Ireland with global talents, fully focused on creating the future we had envisioned: to place human insight at the heart of machine learning and help brands deliver online experiences that meet the needs and expectations of their individual customers.
What is the vision behind Peroptyx?
Our vision is to be the place where brands go to build trust in their on-line experiences. Our mission is to design and deliver the most relevant data quality and model evaluation solutions for machine learning. Our team is born global; coming from Europe, The Middle East, Africa, Asia and the United States. Our customer’s past, present and future all share our tenets.
How difficult was the start and which challenges you had to overcome?
We made many mistakes! Conceptually, machine learning wasn’t widely understood as a discipline, and so we struggled at times to relay the true value of our solution to investors. We prevailed through a combination of experience, relevant examples and metaphors. It also helped that our Chairman, as one of a very select group of Internet Pioneers, was well placed to articulate the significance of AI, Machine Learning and Blockchain as the next true technology megatrends after the internet itself.
Those who understood took a long-term view and came on board. The approach we took was successful for us and we raised over $2 million to bring the company to the next level.
Who is your target audience?
Our target customers are consumer-facing internet platforms and potential unicorns who believe that going ‘straight to ML’ is the most cost-effective and scalable way to deliver truly personalized and authentic user experiences with their online services.
What is the USP of your startup?
Our USP at Peroptyx is that we can rapidly establish a data quality and model performance baseline for any given customer use-case and improve it by sourcing and retaining the ideal expert people to address bias and errors in training data.
To deliver this value, we have deployed a platform that enables us to offer extraordinary levels of global data quality, security and model evaluation capability usually associated with a managed service, but with the speed and economics usually associated with SaaS.
Can you describe your typical workday?
Talking to current and potential customers, executing the next stage of our business plan with the team and completing at least two things that I don’t really want to do.
Where do you see yourself and your startup Peroptyx in five years?
I see us solving harder problems for our customers! Our goal is to grow with them as we help them expand into new markets while avoiding the pitfalls – our past experience in successfully scaling a business to over 100 countries is helpful here.
On day one we had data enhancement and model evaluation capability in 12 markets. This is growing and we expect to reach over 30 new markets across Africa, Asia, and The Caribbean in the next 18 months. We’ll have evaluated our progress at that point and depending on circumstances, will decide where we go next.
We operate on a series of 9-month sprints rather than 5-year plans. Either way, our direction will be determined by and aligned with our Customers’ needs and expectations
What 3 tips would you give to founders?
Love your customers’ problems! Do at least one new thing at least once a week to demonstrate your commitment to solving them with imagination and ingenuity.
If you can afford to, take a long-term view. Success is about laying out your stepping stones, using them to move forward and reusing them to extend your path.
Develop a common language of understanding for your business with your customers, employees, partners & investors.
Thank you Sebastian Kirsch for the Interview
Statements of the author and the interviewee do not necessarily represent the editors and the publisher opinion again.
