In this interview, we had the pleasure of speaking with Amira Dinari and Niti Alekh from Volvo Cars! Amira, Data/ML Engineer, and Niti, Data Scientist/Data Engineer, share their insights on the innovative data-driven projects transforming the automotive industry at Volvo.
In their upcoming presentation at the NDSML Summit 2024, they will speak on Kubeflow for personalization and driving strategy with machine learning. This session will provide a step-by-step guide for building customer segmentation and dynamic communication. It will also cover feedback-driven engagement using Volvo’s in-house ML platform, Abakus.
Hyperight: Can you tell us about your journeys into data science and machine learning? What inspired you to pursue a career in these fields?

Amira Dinari: I am an ML/Data Engineer with a demonstrated history of working with large datasets. I have experience utilizing various technologies and programming languages. I possess skills in Python Programming Language, Machine Learning, Scala, and Data Science.

Niti Alekh: I am a Data Scientist with 10 years of experience in analyzing large data sets and creating data-driven insights using various predictive data models and customer insights.
I specialize in applying statistical inference for real-time scenarios, utilizing different ML frameworks for both predictive and non-predictive algorithms.
Amira & Niti: Our passion for addressing real-world business challenges through the application of machine learning and data analytics has been the cornerstone of our journey toward becoming Data and ML Engineers.
Hyperight: At the NDSML Summit 2024, you will present on Volvo’s in-house ML platform, Abakus. Can you give us an overview of how Abakus was developed?
Amira & Niti: Abakus was designed as an end-to-end machine learning platform, supporting every phase of an ML project, from initial experimentation and prototyping to the deployment of fully productionized models.
Hyperight: Your talk will touch on ML-driven customer segmentation and scoring. How does Abakus help in building these models? Can you share some challenges you’ve faced in this process?
Amira & Niti: We used Abakus during the exploratory data analysis (EDA) phase to evaluate the feasibility of our proof of concept (POC). Our primary challenges revolved around data sourced from different channels and systems. Each source presented a variety of formats and volumes.
Hyperight: In your talk, you will present on transforming ML operations through scalable frameworks. What are some key advantages of scaling ML operations using Kubeflow?
Amira & Niti: The Abakus pipelines significantly streamlined our machine learning operations and enhanced the marketing techniques employed at Volvo Cars.
Hyperight: How do you ensure your ML models remain adaptable to changing customer behavior, particularly regarding segmentation and scoring?
Amira & Niti: We place a strong emphasis on monitoring the performance of our ML models. In addition, we actively collaborate with our business teams to identify and respond to any shifts in consumer behavior.
Hyperight: What role does feedback play in enhancing ML-driven customer experiences at Volvo Cars? How is this feedback integrated into Abakus to improve future predictions and interactions?
Amira & Niti: Feedback data is essential for gaining a comprehensive understanding of customer interactions, enabling us to create more personalized marketing campaigns. We integrate this feedback data into the Abakus pipelines. Here, we refine our customer segmentation based on their interactions with previous campaigns.
Hyperight: What key lessons have you learned in your work with MLOps that would be valuable for engineers looking to scale ML platforms?
Amira & Niti: Start small and fail fast! Not every machine learning project will make it to production. However, the key is to collaborate closely with the business team. MLOps (Machine Learning Operations) isn’t just about the tools; it’s about adapting best practices to fit the unique solutions your team has developed.
Hyperight: Looking ahead, what technologies in ML and data engineering do you believe will have a big impact on the automotive industry in the next few years?
Amira & Niti: Federated learning (FL) is an emerging technology that we believe will significantly improve the efficiency and safety of autonomous driving systems. By enabling AI models to learn from data generated across a fleet of vehicles, FL significantly improves the system’s overall intelligence and adaptability.
Tune into Amira and Niti’s presentation on “Kubeflow for Personalization: Drive Strategy with ML,” at the NDSML Summit 2024. Don’t miss this opportunity to explore the future of machine learning!
Add comment