Resume

• CS 70 Discrete Mathematics & Probability Theory (Prof. Rao, Prof. Ayazifar)
• CS 61A Structure & Interpretation of Computer Programs (Prof. DeNero)
• CS 61B Data Structures & Algorithms (Prof. Hilfinger)
• CS 61C Great Ideas in Computer Architecture (Prof. Garcia)
• CS 170 Efficient Algorithms & Intractable Problems (Prof. Haghtalab, Prof. Wright)
• CS 186 Introduction to Database Systems (Prof. Cheung)
• CS 162 Operating Systems & Systems Programming (Prof. Kubiatowicz)
• CS 189 Introduction to Machine Learning (Prof. Listgarten, Prof. Saremi)

• MATH 1A Calculus I (Prof. Vojta)
• MATH 1B Calculus II (Prof. Paulin)
• MATH 54 Linear Algebra & Differential Equations (Prof. Lott)
• MATH 110 Abstract Linear Algebra (Prof. Frenkel)

In progress:

• CS 182 Deep Neural Networks (Prof. Sahai)
• MATH 113 Abstract Algebra (Prof. Utiralova)

*Courses in bold are favorites of mine!

Improved automatic tuning of hyper- and production parameters for models that are online-trained by lowering the overhead of trial-based experimentation, which is used in their automatic tuning routines. Examples of these models are those that estimate the probability of a user’s click on a given ad on their feed. Tuning these parameters helps minimize both loss and compute-related cost overhead.

This mostly looked like writing typed Python and working with a very wide breadth of internal tooling, including HPC schedulers, database services, and distributed computing frameworks.

I contributed to the Meta Ads platform. Specifically, I worked on a pre-alpha feature related to ad bidding which will help advertisers express granular adjustments to the value they attribute to different audiences based on various user dimensions, and which will be a part of the Ads Manager tool once released. This was full-stack product work.

I was taught large-scale application design and developed my own project with the guidance of two (incredibly didactic and supportive) software engineers. While I barely touched Meta’s codebase (by the program’s design), I did learn a lot about the work of the team I was placed into, which helped WhatsApp transfer and archive data.

The main project I worked on was an iOS task-based project management application I called Process whose main feature was a directed acyclic graph of task-to-task dependencies (as opposed to just a tree or list) on a recursive user interface. It is built on a Firebase-d backend and SwiftUI frontend (in hindsight, should have used Neo4j’s graph DB instead).

I led a team responsible for the development of open-ended IT initiatives in the department of Student Affairs and the management of all workstations and devices used by UC Berkeley Student Affairs groups. Here are some of the things we achieved together during my time there:

• Built infrastructure for and provided 60 iPads to teachers in the ECE Program.
• Created 2 interviewing stations free for all UC Berkeley students at the Career Center.
• Set up an accessible station for CalDining employees with low tech and English literacy.
• Helped Cal Esports do a drop-in replacement of 55 gaming PCs’ hardware and software.

In addition to these collective achievements, in my specific capacity as team lead I was able to:

• Conduct 4 successful hiring panels for positions on my team (including my own position).
• Advise ex-CIO Jenn Stringer and IT leadership as part of the Student Technology Council.
• Contribute to the OneIT initiative instantiating more student programs in Berkeley IT.

After leading the Endpoint Engineering team for about 2 years, I transitioned to a Senior role to enjoy additional time for pursuing my research interests. I focus on supporting my team in additional onboarding as well as other ongoing efforts, such as improving our implementation of mobile device management.

I provided frontline hardware and software assistance to students, and contributed to our standardized procedures which are still used by the 20+ member Student Helpdesk team. This was overall incredibly varying work, and I learned a lot about customer service and got to be very scrappy (in a careful way). Students can sometimes get their computers into truly “wow” states.

No meal is complete without the bread, and no CV is complete without fast food experience. With the help of my comrades, I served food at up to 3 orders per minute. I worked 40 hour weeks while being a full-time high school student, which was exhausting. This is the hardest work I have done in my life, including being a Berkeley student (but excluding 40 degree Celsius soccer practices in 2nd grade).

I own a Rust platform for analyzing sequential games called GamesmanNova. It was originally built as an informed re-architecture of my advisor’s original C software Gamesman.

My recent work has been aimed towards translating encodings of provably perfect strategy in sequential games to representations that are intuitive to humans in a reproducible and abstract fashion by expressing them in terms of observable features of game states. In the past, I have helped with:

• Desigining domain-specfic databse systems and algorithms for analyzing sequential games.
• Administrating the instructional component of our group, teaching 20+ students per semester.
• Solving Five-Field Kono (you can play against the solution here).
• Outreach during CalDay ‘23 and ‘24!

This is the FIRST Robotics Competition group from one of the high schools I went to. It is very close to my heart, because me and my closest friends put in a lot of work for the team. While I still occasionally help out with mentoring, my active involvement has largely ceased because I have gotten increasingly busy. Here are the roles I have held:

• As an official mentor I gave professional and academic guidance to the team.
• As head of the Design Division I coordinated the design of an award-winning robot.
• As a driver I participated in competitions and got to crash our robots a few times.
• As a member I made a lot of friends who went on to become incredible individuals.