Meet Gabriele, Konfir’s Data Scientist

By Tom McAuliffe

Gabriele joined the team in July as a Data Scientist, bringing with him a wealth of experience across a multitude of disciplines, including machine learning, consulting, numerical modelling, big data, parallel computing, optimisation, and software development. Holding a PhD in Numerical Modelling, Gabriele is at the forefront of our initiatives to convert data from candidate-permissioned sources into usable employment verifications, making it easier for verifiers to navigate complex screening processes.

What sets Konfir apart from other places you’ve worked?

It’s really a blend of two key factors: speed of execution and a horizontal culture that places a strong product focus.

In terms of speed, Konfir is all about “learning quickly and failing fast,” which is crucial when dealing with big data and machine learning. We’re always iterating and improving, ensuring that we consistently deliver value to our customers.

As for the culture, it’s incredibly collaborative and horizontal. Everyone is involved in multiple facets of a project, which allows for a diversity of perspectives, cross-functional understanding, and better problem-solving. Every decision is guided by the potential value it brings to our users, rather than just its technical viability.

What have you learned since joining the team?

I quickly realised that the ideal solutions often require a mix of resources, not strictly limited to machine learning. The most critical area I had to accelerate my expertise in was dealing with fuzzy matches, similarity, and ranking of textual data. These have been instrumental in ensuring we can uphold world class accuracy and speed.

Additionally, it’s been invaluable to maintain a close working relationship with the Engineering and Product teams. This collaboration allows me to understand crucial usability constraints like latency and throughput, which directly impact the user experience.

We also experimented with some Large Language Models (LLMs), which although a hot-topic, not a fit for our current dataset and requirements. We found it to be unsuitable for our current needs, mainly due to speed and the types of data we deal with.

What do you expect Data Science at Konfir to look like 12 months from now?

A year from now, I envision Data Science at Konfir transforming into a fully integrated MLOps solution with human-in-the-loop capabilities. Given that Konfir’s product is overwhelmingly data-driven, the ML solution needs to be designed as non-static and retrainable from the start.

The human-in-the-loop component is essential from both an ethical and practical standpoint. Our solutions have real-world impact on people’s lives, so it’s crucial for human expertise to intervene where machine learning alone might fall short.

Beyond that, I’m excited about the prospects of implementing risk scores and diving deeper into employment analytics. I’m particularly interested in tracking career trajectory shifts, like movements from blue-collar to white-collar jobs or from one industry to another. These insights could be invaluable for both our team and our customers.

What guiding principles shape your way of working?

My mantra revolves around the idea that “no data is fixed; all data is flexible.” This underscores the reality that data is an ever-changing landscape, and our models and rules need to adapt accordingly.

While writing parametric rules might offer a quick solution, they are bound to break or become obsolete as data and needs change. On the flip side, relying solely on machine learning models could introduce its own set of issues, such as biases or overfitting. That’s why I advocate for a hybrid approach that combines the best of both worlds. This enables us to be as data-driven as possible while continually learning and adapting to new patterns and challenges.

So, in essence, my way of working is rooted in flexibility, adaptability, and a commitment to blending rule-based logic with machine-learning insights. It’s about finding that sweet spot where technology meets human intuition, allowing us to deliver the most accurate and valuable solutions to our customers.

What do you value most when bringing new talent onboard?

I prioritise individuals who embody a mix of self-initiative and teamwork. Proficiency in Python programming is a foundational requirement; however, beyond that, we appreciate candidates who bring a fresh perspective to the complex issues we tackle.

It’s important to underscore that our projects are rarely purely algorithmic in nature. They often require a cross-functional approach that integrates data engineering with data science, all underpinned by a coherent strategic vision. Therefore, we value team members who not only possess technical acumen but also have the ability to engage with broader, multi-disciplinary challenges that align with Konfir’s strategic goals.

Back to blog home