BrazilAndrei Pascanean on his return to Sia Partners
A year after departing the firm for another adventure, Andrei Pascanean has returned to Sia Partners in its Amsterdam office.
Having obtained a master’s degree at the Erasmus School of Economics, Andrei started his professional career at the international management and technology consultancy in 2020 as a data science consultant. With valuable experience under his belt, Andrei then left Sia Partners to explore an opportunity at a machine learning specialist.
Now back at Sia Partners, Andrei is currently working on two projects simultaneously. For a large retailer in Europe he is supporting a data-driven approach for a new product pricing strategy, meanwhile, at an international client in the energy landscape he is part of a team that is helping the company with enhancing its customer analytics processes.
What did you work on during your first stint at Sia Partners?
During my first year as part of the Data Science practice I worked on a sales analytics project in the industrial B2B sector, a system to forecast solar energy generation and also developed internal products.
For our B2B sales client the goal was to derive key insights from the available data to provide business insights in the short-term. As part of a long-term analytics plan, I also got to do some use-case discovery where I had to combine my data skills with the ability to ask the right business questions.
Understanding the needs of the client, knowing what is and what is not possible in the realm of data science and having some creativity are all skills I learned where important when consulting. Creativity was especially important in a later project where I had to solve the ‘cold-start’ problem for solar energy forecasting.
As part of Sia Partners’ initiative for equality in the workplace, I conducted research on inclusivity in the job market. This resulted in the ‘Diversity & Inclusion Bot’, a Sia Partners product used to analyse job advertisements for inclusivity using machine learning. Our research was published in the Economist, Financial Times and Consultancy.nl.
Other internal work included building machine learning packages and marketing them through training videos. This was a chance for me to learn all about writing clean code for production and effectively distributing your work for others to contribute.
Why did you leave Sia Partners after one year?
As a young professional who had recently graduated, I had the idea of trying out different data science roles. One part of this was experimenting with different domains where data science can be applied, which was the reason I chose for a consulting role.
Another important part for me was understanding what type of consulting would best fit my skills and personality. After having experienced a hybrid of strategy and technical consulting at Sia Partners, I wanted to try out a pure technical consulting role. I felt that early in my career I still had the opportunity to switch roles and decide what I liked best. That is why I switched to a deep technical job where the focus was on implementing machine learning solutions.
What did you learn in the past year outside of Sia Partners?
During my year away, I developed and deployed machine learning models in the energy industry. I was lucky to have the opportunity to work on an end-to-end data project where I got to experience all parts of the machine learning product life cycle. I learned how to build data pipelines, maintain models after training them and deploy models in production for the end-user to interact with.
I was mainly focused on gaining in-depth technical experience through both hands-on client work and interactive weekly training sessions.
The level of complexity in my day-to-day work is something that I was unable to achieve in my first year at Sia Partners. Being in a more technical environment allowed me to get in touch with a lot of different techniques in data engineering, machine learning engineering and production-grade development. I got to learn these skills as part of a team that focused on implementing the client’s machine learning vision as technically efficient as possible.
What was the reason you re-joined Sia Partners?
I enjoyed working in a more technical role, but realised that I did miss being involved in the project lifecycle much earlier. I missed the first step in a data science project, that requires interaction with business stakeholders to translate their business needs into a data solution. Being able to influence the type of solution implemented, from both the business goals that need to be met as well as the technical specifications, is a unique opportunity to create impact.
The challenge of understanding a client’s needs and then proposing and implementing a solution that you yourself defined requires both business and technical skill. These are truly end-to-end projects that go from a user’s problem to whatever implementation required to solve it, creating change along the way.
As part of this journey I also have the opportunity to interact with business stakeholders on management and executive level. Communicating your work’s results effectively to the business side while also working together with technical experts is a unique and fun challenge.
What are you working on right now?
Currently I am working on projects in the retail sector. My first project is to advise a customer-facing energy client from Dubai on implementing customer analytics as part of their new loyalty program. This includes identifying relevant use-cases that can be implemented, architecting the required IT infrastructure and pinpointing the values that our client’s marketing department can obtain.
I also have meetings with managers and executives from the client side who I have to convince of the project’s business value.
I’m also currently working for one of the biggest retail chains in Belgium to help define a new product pricing strategy. As part of this deliverable I get hands-on experience pre-processing data and building models that can be used to make company-wide business decisions. This combination of technical implementation and strategy recommendations is what I feel most comfortable with.
What are you looking forward to further developing?
My main focus in the upcoming year is to develop a more overview-oriented approach to data science solutions. That includes becoming more familiar with architecting solutions in the cloud for a wide range of use-cases. A part of this goal is obtaining certifications with cloud providers, while another part is working on projects where Sia Partners recommends clients how to set up their cloud environment from the ground up.
Of course the end goal is always to deliver value to clients, irrespective of the technical back-end. To do this I want to develop my skills of identifying quick-wins and of translating between the business and the data side.
Finding so-called quick-wins that have minimal technical complexity but can provide immediate value is essential at the start of a project. Equally important is being able to construct more complicated machine learning models when necessary and then explain their usage effectively to business stakeholders with minimal technical understanding. This is what I hope to practice and develop further at Sia Partners.
Do you have tips for people who are looking to apply for a data consulting role?
I would advise potential data consultants to understand the ‘why’ of an analysis more than the ‘how’. I know this sounds very cliché but you need to see that data science tools do not serve a purpose in and of themselves, they are used to extract business value from data.
Start by understanding why you need to perform a certain analysis. Who are you doing this for? What does she or he want to achieve using your analysis? How would your business stakeholders end up using your findings or model?
Often people rush to interpolate missing data, fit the 10 best models for the use-case and sort the performance results by some arbitrary metric. The real world is not a deep learning 101 class where you are handed an assignment with some nicely prepared data. Sometimes a simple SQL query can already solve 50% of the problem, or at least provide a nice baseline to start from.
Providing business value to the end user should always be the goal, no matter how that is achieved from a technical perspective. So always remember why you are building a data pipeline or machine learning model, ask yourself ‘so what?’ at every result.
Alongside his daily job as a consultant, Andrei Pascanean is also a fervent blogger about data science and life in consulting. Follow Andrei’s account on Medium for his insights.
