On Thursday, July 23rd, Correlation One hosted the sixth edition of the C1 Connect Summer Webinar Series: Data Science @ Work x causaLens featuring causaLens Co-Founder and CEO, Darko Matovski.
causaLens is a London based startup on a mission to build truly intelligent machines that go beyond what is possible with current machine learning approaches. Providing an Enterprise platform for businesses in finance, IoT, energy, and telecoms, causaLens helps businesses create accurate and robust predictions, automate data cleaning and searching, discover autonomous models, and stream end-to-end productization.
During the webinar, Darko shared the story behind causaLens explaining that today, machine learning techniques are really good at predicting specific outputs by looking for correlations between varied static inputs. However, the success of these models is subject to humans' ability to curate and update their inputs, leading to models that often fail to account for the dynamic nature of the global economy and the world surrounding us. To even start understanding the world using traditional, or theoretical data science techniques requires building many, many models that must accurately reflect the problems of our day and capture the complexity of our world, all the while constantly adapting them to reflect the environment around us.
"current Machine Learning techniques are based on correlations and they are unable to understand true causal drivers of a change in our world [...] and when the world changes these correlations are not relevant anymore"
Along with his co-founder Maksim Sipos, Darko identified that it took too much time and too many resources to build predictive models that drive business outcomes and that even the best models become obsolete very quickly, sometimes within a day-- making current data science techniques insufficiently scalable. CausaLens was founded to devise and implement Causal AI, teaching machines cause and effect for the first time - a major step towards true Artificial General Intelligence.
"Almost all current machine learning approaches, including AutoML solutions, severely overfit on time-series problems and therefore fail to unlock the true potential of AI for the enterprise," says Darko. During the webinar, Darko helped our audience understand the difference between current AI and ML practices and true Causal AI as well as the research and resources that he relies on to stay up-to-speed with the latest advances in data science and AI research.
Darko also shared his data science career journey, his vision for the future of data science, and how we can unlock the full potential of data science in the real world. If you're curious to learn more, including Darko's answers to following the questions from our audience, you can watch the Webinar below:
-What are some of the key breakthroughs that have allowed for the transition from correlative to causal predictions?
-How would you advise a candidate to determine if a data science role is a 'fit' for his/her desired level of sophistication?
-causaLens solves complex, real world problems for companies in varied industries. Can you share your approach to developing reusable models and solutions?
-What makes a candidate stand out during causaLens's candidate experience?
If you are interested in opportunities at causaLens, please apply to C1 Connect here.
About Data Science @ Work
There is a transparency problem in the data talent market.
At C1 we work with thousands of data scientists, data analysts, and data engineers from around the world, and we often hear from job candidates that they are unsure how to evaluate different data career paths, do not know what skills they should focus on developing, and need some guidance on how to find their next data science job.
Across industries, companies are challenged to define the difference between a great data scientist, data analyst, and data engineer on job descriptions. This makes it difficult for candidates to understand what their day-to-day responsibilities will be, how certain jobs will impact their career trajectories, and how common job titles like 'data scientist' differ from one company to another.
This lack of transparency leads to a huge waste of time for both candidates and companies. Candidates adopt 'spray and pray' job application strategies, applying to hundreds of roles that have 'data' in their title. Talent teams are then forced to search through thousands of resumes to find great candidates who then must be triaged to the appropriate role search. Oftentimes, the interview process uncovers that though a candidate is an excellent data scientist, her goals and skills do not align with the role. This wastes the time of the applicant and Senior Data Scientists responsible for conducting late stage technical interviews.
We launched the C1 Connect Data Science @ Work webinar series to break down the communication barriers between hirers and the world's best data scientists, data analysts, and data engineers. Each week, our C1 Connect community is invited to hear directly from data leaders who share background on their career journeys, what working in their industry means practically for data professionals, and some tips for navigating the job search (and if applicable, how they can pursue opportunities with their teams).
After each session, candidates are invited to raise their hand for feature opportunities on C1 Connect by sharing their C1 Connect Datafolios- brief profiles designed to communicate the skills, roles, aspirations, and project work specifically for data professionals. Using C1 Connect's Talent Match Algorithm, we pass on qualified candidates who fit the profile for active opportunities to the proper next steps in the candidate selection process.