Alectio’s story begins long before its incorporation in January 2019.
It began when our founder, Dr. Jennifer Prendki, transitioned into the industry after a PhD and a PostDoc in Particle Physics. While she had no issue applying the data analytics skills she had developed as an experimental researcher, she couldn’t grasp why the Data Industry was so reckless about the way they were managing their data. In academia, the computational resources that were available to her were extremely limited (you literally had to request time on the university’s server to run statistical models), and the way other data scientists were training models felt wasteful, to say the least. Regardless, she tried to get used to it, and did not mention it for years.
It wasn’t until 2016 when she was put in charge of a large team of data scientists with a very limited budget and limited computational resources once again, that she felt her early-career instincts were on point. Tired of constantly begging for budgetary extensions for data labeling and servers, she started researching ways to train models with less data and became one of the earliest proponents of what would later become the Data-Centric AI proposition.
After years of evangelizing MLOps best practices, Active Learning and other data-efficient approaches to Machine Learning as a way to sustain progress in AI, she eventually took her mission to the next level, first by becoming the VP of Machine Learning at Figure Eight (previously Crowdflower), a pioneer in Data Preparation, and eventually started Alectio to provide solutions for any ML scientist to build models with less data, and help prevent the next AI Winter.
When she founded the company, Jennifer’s focus was all on Data Curation, and specifically, on Active Learning, so she knew that the company’s name had to include AL. So one evening, she proceeded to write close to 120 different names containing AL, and decided to make a decision in the morning. But when she woke up, the only name she could think about was one she had not included in her list. It was Alectio, AL for Active Learning, and lectio, a latin term for “selection”, which seemed suited to the problem she was solving. A cycle to represent the cyclical nature of Machine Learning, and we were good to go!
What’s the Story behind the DataPrepOps trademark?
At the beginning of January 2020, our founder was invited to give a talk at TWIMLCon. She sent a proposal to the organizers, and the suggested topic (smart data preparation) was immediately accepted. However, she was asked to find a more compelling title. DataPrepOps immediately popped into her mind almost instantly as the most natural way of explaining the company’s business model. The terms caught up almost immediately within the industry, and luckily she had had the presence of mind to trademark “The DataPrepOps Company” by the time the term started becoming more mainstream. So nowadays, we are Alectio – The DataPrepOps Company.
What’s with the pencil style of the website?
We are pioneering a new concept, a new field, and countless new technologies for which no open source libraries exist, and often no paper was ever written. We are about innovation. So the least we can say is that our products and algorithms are a constant work in progress. We are a work in progress, and none of the work we do is ever final. We’re researchers and experimentalists and are not ashamed to scrape everything and start over from scratch. Hence the pencil style.