Hi! I'm Neil,
and I'm a recent MIT grad working on code & policy.
more about:
me
work
projects
resume
Hey! I'm Neil Malur, and I just got my bachelor's degree in Computer Science and Engineering from MIT! 🎓 Right now, I'm working at the Indian Institute of Science's ARTPARK in machine learning research for heatwave detection before I start full-time this fall as a Software Engineer at Meta. I think a lot about how to build better systems in the world, and I try to use code to solve problems on the way there. In my free time, you can probably find me dancing 🕺, chatting with someone random off the street 👯, or figuring out how to fix my bike pedal 🚲 (again), but feel free to check out some of my work on the other tabs!
My most recent work has been at Nominal, a startup that focuses on automating the hardware test and validation lifecycle. At Nominal, I focused most on our data systems and processing, particularly on speeding up our file ingest pipeline, adding additional REST API compatibility, and augmenting our data computation offerings. Before Nominal, I got a taste of policy at the United States Department of Commerce, where I used machine learning to help shape macroeconomic policy tailored to alleviate geographic inequality. At the DOC, I also worked on developing an in-depth case study of the metro Charlotte area based on original economic analysis.
I've also found myself teaching a lot lately. This year, I co-taught a machine learning course to graduate students at La Universidad de Ingeniería y Tecnología in Peru, with a focus on combining theory with actionable projects and startups. Last year, I taught web design and machine learning to undergraduate students and young professionals in Jordan with Injaz :)
My current interests are in computer security and natural language processing, and I'm especially excited to work on projects that apply them to a policy context.
Here are brief descriptions of some projects I've worked on in the last 2-3 years!
DIE-BERT- As a final project for a Natural Language Processing course, we augmented BERT to accept prosody-related information, including relative spoken word length. We used forced alignment to gather this data from speech, modified BERT's input layer to accept our additional encoding, and fine-tuned the model on the results.
- With MIT's Climate and Sustainability Consortium, we designed and implemented a reinforcement learning-based architecture for polymer property prediction, particularly ensuring that crystallization would create valid conformations. We used a novel framework and strategies to reduce the search space of the folding of very long polymer chains, and our work was presented at an ICML workshop!
- For a final project in my Systems Security class, I secured a social media site by augmenting it to support WebAuthn 2.0 compliant security key login and to consistently fetch certificates from an ACME server.
- As a Social and Ethical Responsibilities of Computing Scholar at MIT, I worked with a lab to produce a timeline and analysis of the spread of COVID-19 misinformation using natural language processing on scraped social media information. We used both more analog NLP methods as well as pipelined LLM tagging on primarily Reddit and Twitter data.
- As a final project for a Computer Networks course, we designed and implemented an overlay network and congestion control mechanism to optimize multi-user file download on a shared network. Using partitioning of the file, consistent re-evaluation of link usage, and a custom pathing protocol, we achieved the most consistent and one of the highest performances for large files.
- To compete in MIT's Web Lab monthlong competition, we created a React-based site with hooks to generate recipes based on user-inputted ingredients. Once the recipe was created, the site would launch a custom game to go find all the ingredients!