The 2021 Bloomberg Team
Sprint 1: It’s time for Boarding!
The Bloomberg Team
At the beginning of 2021, an all-star team of Masters of Human-Computer Interaction students at Carnegie Mellon was formed. Our team has a wide variety of disciplines and backgrounds but are all aligned by a passion for learning about Machine Learning and Artificial Intelligence applications:
Jennifer Park — LinkedIn
Jennifer has a background in product management and development. She has previous experience working on consumer-facing products with a lens for user experience and a data-driven approach.
Favorite Human-AI Duo: Wall-E and Humans
Monica Chang — LinkedIn
Monica comes from an apparel design background. Her time in the fashion industry allows her to think systematically about product creation and design with a human-centered approach.
Favorite Human-AI Duo: Baymax and Hiro
Nathan Jen — LinkedIn
Nathan is a recent 2020 graduate from CMU with a bachelor's in Information Systems. He has previously interned at Deloitte as a technology analyst and has experience prototyping web and mobile applications.
Favorite Human-AI Duo: Sam Bell and GERTY
Nitya Devireddy — LinkedIn
Nitya has a background in neuroscience and cognitive psychology. She has previous experience prototyping and designing interactive games and activities as a cognitive skills trainer in a neuropsychology clinic.
Favorite Human-AI Duo: Iron Man and Jarvis
Raajat Gupta — LinkedIn
Raajat has a background in strategy consulting working at Monitor Deloitte. He has extensive experience helping Fortune 500 clients in the financial services space incorporate emerging technologies into their long-term strategies.
Favorite Human-AI Duo: Luke Skywalker and R2-D2
Our team has come together with the mission to optimize the human computational system interaction for Bloomberg’s Global Data team. We were instantly excited about the project’s direction and focus to facilitate human interactions into a machine learning system.
Exploring the Problem Space
As most of our team does not have much experience in the world of AI/ML, we looked to available resources in preparation for meeting with our client. Since the initial prompt that we had received was quite vague, we did our best to cover our grounds. Resources such as Youtube, research papers, and speaking with faculty at CMU have been great in kicking off learning more about Bloomberg and the ML domain. We wanted to divide our time to cover as much breadth and depth as possible before discussing with the Bloomberg team. Subsequently, we divided our research across multiple domains — Bloomberg’s background, the model development lifecycle (MLDC), human-AI collaboration, and industry context. Internally, we had weekly discussions around the machine learning papers and resources we had read which facilitated mutual understanding of deep technical methodologies.
A major highlight for us was the opportunity to talk to our contacts for the first time via Zoom. We contacted the previous alumni from the CMU MHCI Bloomberg Capstone teams to learn about their experiences and gain a holistic understanding of the past projects. Furthermore, we reached out to industry professionals such as Melody Yin (current Bloomberg employee) and Mary Beth Kery (a Ph.D. student at CMU’s HCII and previous Bloomberg intern). From this, we gathered important insights about the current processes and the impact of machine learning on business processes at Bloomberg.
We lived and breathed machine learning! But we are just getting started…
Preparing for Kick-Off
The first item on the agenda was devising a plan for our kick-off meeting with our clients! For our kick-off, we researched design exercises across different companies (e.g IBM, Airbnb) and adapted our own activities based on our individual project needs. This gave us a chance to deepen our understanding of the project but also meet our mentors, Anna Abovyan and Derek Wahila, who gave us valuable feedback during the formulation process.
The goals we wanted to achieve by the end of the kick-off included:
- Develop a clear understanding of our project scope
- Understand the workflow at Bloomberg
- Develop a core problem statement to provide guidance on where to begin research
- Confidence, from both Bloomberg and our team!
We separated our kick-off into two days: day one primarily focused on defining the current state and day two reflected on the future state of our problem. Additionally, we provided some pre-work for the Bloomberg team to get them to start brainstorming about the possibilities!
To increase the level of engagement and foster a sense of fun for our remote meeting, we created a unifying theme in Miro of a cruise ship to drive the team forward to our final destination. Our cruise ship “stopped” at each planned activity on our agenda towards our final destination (the project plan).
It’s Time for Boarding: Kick-Off Day 1
Our first kick-off day was finally here! After days of preparation internally with our team and cross-functional feedback, we kicked off our first client meeting with Bloomberg. Before meeting them, we sent out pre-work that included an icebreaker along with the ‘Hopes and Fears’ activity. Not only did this allow us to get to know their team but also provided insights on some of their concerns and measurements of success. The bulk of our kick-off was focused on facilitating deeper conversations between the teams to fill in the gaps of our understanding and dive deeper into the various pain points. These exercises helped illuminate the core problem areas and the focus of the scope for the rest of the quarter.
Aligned Theme after Day 1:
Users need their mental models and contexts to be incorporated as they work through model development.
Overall, it was exciting to meet every team member from Bloomberg and learn more about opportunities to make an impact on their team.
But wait there’s more: Research Planning and Preparation
After our ✨super✨ successful Day 1 kick-off, we were excited to dive deeper into more research. As a team, we devised a plan in order to divide and conquer. Now that we had a better idea of the scope of the project, we were able to devise more thoughtful primary and secondary research goals. In preparation for Spring 2, we have started to draft the research protocols and design workshops so we can hit the ground running once we have contacts from Bloomberg. This does not mean the research stops! We are charging straight ahead with more secondary research. What we are trying to uncover:
- What are the current solutions in the market and what practices do they use?
- What are some HCI practices for making ML more accessible and understandable?
- What are the different ways of creating “human-in-the-loop” setups?
Our first two weeks have been a jam-packed learning experience! As we move into the second kick-off meeting, the team is ready to move away from the setup phase and dive into the beginnings of compelling generative research. We are fortunate to have the opportunity to work with the Bloomberg team and are intrigued to submerge ourselves into the world of ML.
