What does an Applied Machine Learning Engineer do?

The Machine Learning team use machine learning and advanced data analytics to support progress for patients and research partners.

Researchers and clinicians detect conditions by using different kinds of datasets including multiomics, free text, imaging and other types of clinical data. These scientific advances help to deliver a range of benefits to patients, from diagnosis to predicting treatment responses.

In this blog, we spoke to Andreia Rogerio, Applied Machine Learning Engineer, about her professional journey and the potential of machine learning in healthcare.

Working in machine learning at Genomics England

In my role as an Applied Machine Learning Engineer at Genomics England, I take on a variety of responsibilities. My main task is to bring machine learning (ML) into practice to help different teams solve complex problems.

I work across the entire machine learning lifecycle—from data ingestion and processing, through data analysis and modelling, to ultimately extracting meaningful insights from what the models reveal.

I’m part of a team of 3 ML engineers who collaborate with several teams at Genomics England, since joining, I have worked with the Rare Disease, the Interpretation Platform, and the Multimodal teams.

My main focus is on modelling genomic data, but I have also worked with pathology reports and structured clinical data.

On a typical day, I may have several meetings to discuss my projects, but most of my time is dedicated to coding. I primarily use Python, writing both software-engineering-style code for large dataset processing pipelines, and data science code for analysis and insights.

The journey to bioinformatics

I have always had many interests, but growing up I was particularly fascinated by this idea of finding the “logic” in things.

I studied bioengineering for four years, but after some experience in bioinformatics internships—and a few close calls with lab fires — I decided to pivot to a master's degree in computer science.

When choosing the courses in this master’s, most of the names weren’t familiar to me, so I chose them based on their descriptions. It turned out I had chosen the entire ML curriculum, which became my specialisation.

I started as a junior data scientist in a consultancy, which gave me solid ML experience. Then A few years ago, I saw that Genomics England was forming an ML team and the job opening was an ideal match for my interests.

How does my work impact patients and participants?

Since I work across several teams, the impact of my work varies with each project. I recently completed a study on several in-silico variant effect predictors, which have the potential to enhance our bioinformatics pipelines.

By improving the accuracy and speed of variant identification, we are helping clinical scientists to reach diagnoses more efficiently for patients with rare conditions, even flagging variants that might otherwise be missed or deprioritised.

With the latest advances in large language models we are finding ways to make use of free-text documents like pathology reports. These reports contain crucial information not found elsewhere, but they also hold private patient details (like names and addresses), so we cannot share them with researchers.

To address this, we have developed a ML pipeline that detects private data and removes it from the report. We can now share these resources more widely, which allows researchers to extract insightful conclusions.

For example, knowing these test results may help them understand how to group individuals with cancer to tailor their treatment according to their specific needs.

What makes me passionate about my role?

I really enjoy working with the power of computers by my side. Being able to build something from scratch or have the computer run overnight is not something you can do in a wet lab.

What excites me most is ML's potential in managing and learning from big data. We are only just beginning to tap into this power in healthcare, and I am motivated by the chance to bring AI-driven insights to personalised medicine.

There is so much we don’t yet know about the genome — perhaps our models can help unlock these mysteries.

What’s up next for me?

At Genomics England, I have come to understand the considerable gap between AI and healthcare compared to other industries.

My focus now is to bridge this gap by finding ways to mentor others, sharing my knowledge in programming with healthcare professionals and genomics with computer scientists alike.