Data Science Team: A Mixture of Talents That Creates The Magic
Over the years, I have read many articles that highlighted the initiation of data science projects in corporate companies. Some of them are successful while many of them went down into the rabbit hole. The trend is also similar in hiring data scientist talents, the title has become so popular, yet many companies fail to gain benefit out of the hiring. The article will share how it is implemented in Telkomsel quite smoothly and emphasize the importance of having the right mix of talents and a clear vision, to reap the benefit out of our data science investment.
Data Science Use Case: it’s not a one-man show
The media somewhat exaggerates the role of a data scientist. A data scientist is frequently expected to act as a superhero that can save the company from losing their revenue or a single person perform magic to uplift the company’s financial performance. Well, part of it might be true but it was never about the role and definitely not a one-man show. A data science project is a collaborative work between a team of data-competent talents, clear vision by leaders, and rigorous implementation.
In Telkomsel, common data science use case is delivered by a team of these technical talents:
Data Engineer
The ones that understand the most of Telkomsel dataset, volume, source system-related and strong understanding and experience in these languages: Python and Pyspark. A data engineer also needs to understand data statistics to perform data quality checking and measurement upon the dataset they are ingesting and the pipelines they are building. Data Engineer nowadays is also expected to have the ability to analyze data and to perform feature engineering. A stable data pipeline is an output expected from data engineers.
Data Scientist
The one that performs the magic needs to be strong in statistical approach and methodologies as well as to be experienced in Python & Pyspark. The responsibility is not limited to making predictive models, but also to perform exploratory data analysis (EDA) and to understand the dataset to create solutions in accordance with the business objectives. The model produced needs to meet certain model quality thresholds and agreed upon among all team members and business users. A data scientist is also expected to be able to communicate and to visualize their work results for the team to easily digest.
Machine Learning Engineer
Considering the huge amount of dataset handled by Telkomsel daily, the ML engineer is by far the hardest job in the data science activities from operational perspective. The person needs to be able to scale up the solution made by data scientists and still meet the acceptable processing time and resource constraints for each of the ML model scoring. ML engineers also need to have the capability of deploying the solution as batch and/or real-time through API. It is common to have an MLE that is strong in data platform skills as well.
Analytics Translator
From the outside, the role might seem similar to the role of business analyst and/or project manager. But it is not: an analytics translator needs to have an understanding on fundamental data engineering, statistical methodologies, ETL concepts, performing experiments, drawing conclusion and insight from the experiment. Furthermore, the analytics translator must be able to push through the experiments into implementation. And most importantly, an analytics translator needs to be able to lead and guide the development of analytic use cases to meet business objectives.
Although Telkomsel has many technical talents, the successful implementation of data science use cases is only possible because of the support and trust from the senior leaders. Data science is an iterative and experimental effort, where one winning initiative might come up from 10 unsuccessful experiments. It was the trust from senior leaders that pushed through the experiments leading to successful implementation. The senior leaders’ support prevents data science use cases from going into the rabbit hole in our company. This trust accelerates cultural and operational shifts in Telkomsel which is currently undergoing a digital transformation.