Meet Hannah Howell, data scientist and principle investigator on the Artificial Intelligence IRAD Programme on our United Kingdom team, as she discusses her history with hacking and her current coding project using human machine teaming.
What is the most enjoyable part of being a data scientist?
Getting so absorbed into reading and analysing the code that I lose track of time. I have recently been given the responsibility of managing my own team, and I am enjoying the challenges this brings. Computers can be very clever, but when important decisions are made a human still needs to be involved. What computers can do very well is assist and quickly provide an analyst with relevant information. I am working on Human Machine Teaming, using machines to help humans prioritise tasks.
What skills and tools do you use on a daily basis?
I use Python and Git every day. The main skills I use are problem solving and thinking logically to try and understand where the code might be going wrong.
How is Human Machine Teaming used currently?
Human Machine Teaming is already used in a number of areas. A very simple example would be in a call center where a computer will ask a series of questions and use voice recognition to ensure you are directed to the right operator. A more complex example is automatically calculating measurements on heart scans so that doctors can quickly analyse scans and diagnose patients. Our research aims to similarly triage data and prioritise what a human analyst sees.
What got you interested in artificial intelligence and machine learning at a young age?
How long would it take for everyone in the world to have a ginger beer plant if they can be split in half every week? Trying to figure that out is one of my favorite childhood memories. The equation 2^n = 7 billion seems obvious now, but at age eight it was fascinating, and the first time I remember being really excited about a problem. As I grew up, I continued to be interested in these types of questions. But it wasn't until studying math at university that I discovered that computers could be used to help answer these questions much more quickly. I took courses in MATLAB and a parallel programming course in Fortran, which opened my eyes to the possibilities that computers offered.
You took part in advanced robotics competitions. How did that shape your career?
I had the opportunity to enter Hackspace, a six-month hobby competition for graduates and apprentices that brings together teams from different engineering disciplines to create a robot that can complete a series of tasks. In the first year, the challenge was creating a robot that could follow a line and solve a maze; my team came in fourth place. We were all fresh graduates and made a lot of rookie errors, which resulted in a lot of the software functionality not being integrated into the robot in time for the competition.
Determined that we could do better, we entered again the following year, designing a 2D plotter that could draw pictures from scalable vector graphics (SVG) files, draw a line down a path and solve a dot-to-dot puzzle using computer vision. We had a basic prototype working within four weeks that could draw a rather wonky circle. This approach meant we could be confident that the software we were writing would work on the final design.
Because of the experience Hackspace gave me, I was invited to be part of the team to design and create Gromitronic, an animated and interactive Gromit designed for the Gromit Unleashed 2 sculpture trail in Bristol, UK. The trail raised over £2 million for Bristol Children's Hospital with over one million people visiting the sculptures over the summer.
University taught me how to write quick and hacky code to solve an immediate problem. However, since working as a data scientist, I learned that code needs to be more robust and sustainable. I was introduced to code reviews, version control, good testing practices and coding standards.
After working in a small team on a well-maintained project for over two years, I decided I needed a new challenge and moved to a legacy project with over four million lines of code. This project showed me a different side to software development. Rather than creating anything new, I was deep diving into other people's code to find bugs and optimize processes. Whenever I am tempted to take a shortcut, I remember the legacy project and it never fails to persuade me to do things properly.
How did becoming a data scientist at Northrop Grumman change things?
After moving to Northrop Grumman, my work and interests have now moved toward machine learning and AI. I am working on projects that detect fake news and using computer vision to help analysts prioritise their tasks.
I had the opportunity to be a part of the Northrop Grumman team that competed in a Hackathon organised by the Defence Science and Technology Laboratory. The challenge of the hack was to map wildfires using a swarm of drones inside the drone simulation environment. We developed a strategy that divided the search area into a grid and a central manager that assigned drones new tasks based on their priority.
I have also attended a Women in AI networking dinner, which featured lightning talks from companies such as Spotify and BBC on the impact and ethics of using AI. This was a great opportunity to meet other women in the AI community.
I love coding; it will always be a tool to answer a question or create something cool. So when people say sitting at a computer all day must be so boring, they can't see the bigger picture of the robot or AI I am working on.
Northrop Grumman provides exciting career opportunities for people with a wide variety of interests, talents, and skills, including data scientists. Click here to search for a career in the United Kingdom.
Return to Life at Northrop Grumman home page.
Meet Northrop Grumman employees working in the UK