Sounds like you learned a lot during the summer! Are there any other elements that stuck out to you about how the nonprofit environment differs from a research environment?

It's really cool how you connected your Laidlaw work to your other pursuits, like your show on WKCR! Have you continued coverage of criminal justice issues on the show by interviewing other subject matter experts?

That sounds super interesting - was there a specific seminar that you attended that you found particularly insightful? And, how closely coupled were those seminars to your work - did they give you ideas for any new avenues of research to pursue?

That must have been challenging - language barriers were something I thought a lot about when considering going abroad. Did you find it easier to manage later on in your project when you had spent more time in Paris?

That must have been a great experience - what was the talk at Chiba University like?

That sounds very interesting - how did you get involved with the Square One project originally? Were there any issues that you focused on in particular over the course of the summer?

Apologies for the late post - I had to get my responses cleared beforehand which delayed my submission. 

Week 1

Q: If your project this summer differs from your project last summer, has last summer’s project influenced your project this year, and if so how?  If your project is different, what tools have you developed to help you work on this project?

A: Last summer, I worked on using computational and machine learning techniques to analyze both fMRI (brain imaging) and text data to contribute to a memory study. My project this summer is focused on helping make tools that can automate the process of triage, enabling first responders to respond to crises quicker and more effectively. While my research this year differs - I'm working with health-related data, and analyzing images and videos rather than text or MRIs - I'm using the same kind of quantitative and software engineering skills to develop algorithms and pipelines that can be used in a variety of contexts. Last summer was helpful for familiarizing myself with tools like Jupyter, working on a server, and using Linux/Bash alongside Python, all skills that I am using for my current project as well.

Week 2:

Q: If your project connects with your research from last summer, explain the ways in which it picks up on themes, issues, or questions that are important to you. How are you expanding on your project from last summer? How is your understanding of this topic evolving?

A: My project connects with some of the research I did last summer in terms of the techniques and processes that I am currently using. Part of my research last year involved using "foundation models" (very powerful, general purpose algorithms) to help with analyzing language; this summer, in order to help create technologies necessary for automated triage, I utilized other cutting-edge foundation models (such as Meta's Segment Anything model) that were attuned to image processing. My skills from last summer in creating analysis pipelines for applying pre-trained foundation models to new data (and understanding the steps of data preprocessing, code adaptation, etc) were crucial for this process.

 I also have an evolving understanding of how triage works and what the constraints would be in a computational setting. Whereas there are many things a doctor can focus on immediately (measuring heart rate, note down injuries, etc) it is much harder for a robot to perform those tasks. There are workaround measures that are themselves areas of research (e.g., identifying heart rate from video of a person), but the process of working on this problem has provided me with a deeper understanding of what those second-order steps are.

Over these past two summers, I feel I have been exploring different types of knowledge - knowledge conveyed by text, by videos, by images, by time series data - and how we can analyze those types of knowledge in concert with each other to build new mental models and tools that can help build our understanding of the brain or help doctors treat patients.

Week 3

Q: What does a typical day look like this summer? Aside from a narrative description, upload a photo, video and/or other media submission!

A: The days tend to fluctuate a bit (especially depending on whether I'm in person or remote), but typically I wake up at 7, take the train and then a shuttle from DC to Laurel, MD, and start my workday. I'll work on my code, and attend meetings on the project. Often these meetings will give me a better picture of other areas of the project (for example, other types of triage approaches that are possible) and the overall topic. After lunch and some more work (interacting with some of the other interns as well), I'll head over to the makerspace to work on the intern project for our sector - essentially a separate project interns across teams can work on together. This year's intern project is making a small drone (like one you might buy on Amazon and fly) from scratch and programming it to fly autonomously. It's a super interesting project (involving hardware, software, and AI) and I've been learning a lot from it! After that, I'll head back home, grab dinner with a friend, and relax.

Here's a picture of the drone that we built (after a crash during a test run):

https://photos.app.goo.gl/Qp2EnNTv89o3gVMX7

Week 4

Q: What challenges and/or difficulties have you encountered and how did you go about resolving them? Speak to a specific challenge you have encountered and some of the ways that you tackled the problem.

A: I've encountered a number of challenges trying to implement my project, due to my lack of familiarity with some of the techniques and approaches that we are using (related to computer vision) as well as my unfamiliarity with the general area of triage and first aid. I've tackled them both by consulting online guides and resources, which are abundant and very helpful, and also by talking to the employees I'm working with to see what they know about the issue and if they have any advice for approaching the problem.

One challenge I encountered was while I was working on my analyses. One type of image analysis is an image segmentation, where you highlight relevant areas of an image that you want to focus on (i.e., segment the area of focus). There are different algorithms that exist for performing these segmentations, where you give the algorithm some kind of "prompt" so it knows what to focus on. I was having issues with using one kind of prompt - a keypoint prompt - where you place dots on the rough areas that you want to focus on. This prompt was not granular enough, and selected elements that I did not want to select (for example, it would select a subject's whole body, instead of just their arm, because it couldn't tell that I was trying to select their arm and not their entire body). To resolve this issue, I talked with the other employees about it and had a one on one meeting with someone who was very experienced with using this type of model, who walked me through some of the alternative prompt methods we could explore. I ended up settling on a bounding box approach, where you draw a box around the desired area instead. This has the benefit of telling the model both what you want to select and what you do not want to select. In general, I have utilized and practiced a lot of my communication skills, which Laidlaw has emphasized throughout my two summers.

Week 5

Q: What new skills and/or knowledge have you gained from your summer experience?

A: I have gained a lot of skills and knowledge from this summer. As mentioned in some of my prior reflections, I gained technical skills, such as performing new types of analyses and generating new software workflows that combine analyses together to automate long and repetitive processes. I also gained technical knowledge, such as how cutting-edge image analysis (computer vision) algorithms work, as well as learning about the different considerations that go into triage and first aid response, especially in disaster situations. Lastly, I exercised and gained soft skills, such as collaboration and communication. Working on a project with a larger team meant working with several different people and communicating what I was working on and struggling with, as well as proactively reaching out to others for assistance and guidance. This enabled me to learn a lot more than if I was just teaching myself, and it ensured that everyone's work was complementing each other's rather than duplicating efforts. I also got to know the employees and interns that I was working with, which was helpful for forming connections and learning more about other people's professional journeys.

Week 6:

Q: For your final post, upload a video presentation or create a written or photographic narrative in which you discuss your project.

A: https://drive.google.com/file/d/1M5DrMegpgxqRVRJuxtcgZtd32fa_ClQJ/view?usp=sharing