Saturday March 20th

A very long pause week

This week was Pause week

  • Apparently, this week was supposed to be a week where we would be able to rest, or no new lessons would be taught, or something to that extent. We still ended up covering some Quantum Cryptography stuff, so I ended up going to class anyways, because the class is awesome, and the teacher is fun and engaging!
  • I ended up doing a lot of work this week, including reviewing a Journal article, and reviewing 5 papers for two separate conferences, as well as going to a training session for a third I haven’t received assignments for yet.
  • I also met with several recruiters, attended several recruiting events, with some more this week, some of which are already for Summer 2022.
  • I also wrapped up a truly amazing class, where I scribed for one lesson. The class was specifically on Information Theory as it related to blockchain, and we looked at different protocols and their tradeoffs, among other things. We also looked at opportunities for new protocols that have not currently been explored, and that was fascinating. We also had a guest lecture by Bram Cohen, the inventor of BitTorrent, who is currently working on Chia. I learned a lot from that class, and also attended several seminars and other affiliated events with that group, and made new friends. I was genuinely surprised when people from the class were passing on opportunities, chatting and just being friendly, and I looked forward to every single class! I joked this week that if I were a Sesame Street character, the word of this week would be Ourobouros. There are a lot of references to that protocol (or derivatives of that protocol) this week. My school doesn’t have a formal Information Theory department, and in fact, I was able to convince a BioEngineering professor who has taught the class before to teach it in Fall, so I guess that’s the next step for me.


  • Information Theory isn’t new for me. I came from film, and was a member of SMPTE, so I’m very familiar with compression and standards. In fact, I’m a member (for several years) of a few groups focused on research in that area, which is very random (and is where I heard about the class I audited, which was based in Silicon Valley). It’s like I was a part of a community without knowing that it was a field.
  • Lately, a lot of this community is involved in quite a few things; one is (obviously) blockchain, and then there are others like distributed and quantum information data compression and genomic compression and of course, video compression and streaming. It’s all fascinating stuff. I think there is this wonderful intersection that blends what I’m doing with that community, so it’s been pretty fun, and I’ve definitely seen a lot of familiar faces in that area.
  • I’m also working with my Google Research mentor, who has been amazing support for me! I forgot to mention that I helped review someone’s dissertation this week, too! It is a really fascinating topic, and I’m really thankful to have been able to help!

Privacy and Crypto

  • I went to another event this week where we discussed Universal IDs and what that would mean for Privacy, which was interesting, especially coming from another country. I wonder if someone who has been born and grew up in one particular country (say, the United States), finds the idea of a Universal ID more compelling than say, someone who hopped around various countries? I quite like that my life in my country of birth has a degree of separation from my life in the United States, to be honest.
  • I’m taking a Cryptography class with a professor who I’ve absolutely enjoyed class with, and I plan to take more of their classes. They even gave me a paper to work through some stuff, so I’m ramping up on that as well. I’ve been writing quite a fair amount of SageMaths, which is pretty similar in many ways to Python, but is more awesome for the out of the box functions for evaluating elliptic curves and that sort of thing. I mentioned in a previous post that I also was part of a SageMaths workshop in my first semester of school (partially because I was very alone and felt really alienated, and mostly because I was curious and excited about the workshop!) where we worked on Arithmetic Dynamic functions for the software, and I ended up contributing to a grant proposal paper as a core member with mathematicians from that group.


  • I’ve once again signed up for more R workshops, and am enjoying them a lot! I signed up for one today specifically on computationally intensive methods and hierarchical models, which should be pretty fun. I was caught off-guard at an event this week where I thought I was going to a social, but was grilled on some questions instead, and they caught me off-guard because I’d been doing cryptography work all week in Sage, and I was asked a bunch of statistics questions. What’s interesting is being at the intersection of those fields; things like Privacy-Preserving Machine Learning put you in that space. It means that you are constantly going back and forth between the encryption side of things, and statistics (unless you are focusing specifically on a mechanism that is data or statistics-focused, or one that is not; great examples are Differential Privacy, which is data-dependent, and say, homomorphic encryption, which isn’t). It’s pretty rough, and I took a nap in the afternoon and got up just in time to see the sun set. I can’t choose, because I like manipulating data, but I’m equally into encryption and privacy. I am not as much into AI for AI’s sake as some of my peers (I get pretty bored in AI-only spaces), but many people working in privacy and encryption aren’t necessarily into data. So I’m in this weird middle, and I can’t help bouncing back and forth. Statistics doesn’t quite describe it, either; encryption and privacy people who work with data are more systems people, than they are solely statistics people. It’s not data analysis; it’s more like distributed systems analysis or something. So you can imagine that “fairness” for someone from a pure stats background has a very different meaning compared to someone from a systems background / way of thinking. In the latter, something like resource allocation is very obvious, in a way that may not be obvious to someone from solely data-centric background. The intuition is just very different.
  • Shafi gave a brilliant talk at NeurIPS 2020 where she spoke about doing ML from a cryptography background. If your foundation is different, you think very differently, as each field’s foundation shapes how you think about problems. It’s interesting because I was chatting with my Google Research mentor today about how the data people always shoot for finding a dataset to do analysis on first. It’s very strange. But it’s by a formulation of a habit of their field and training. They’re constantly thinking “what datasets should we use to evaluate our hypothesis on?”, which can be a bit strange if you come from another field.


  • One of my biggest frustrations when I was learning about tech was meeting people who didn’t care about hardware; only software. To me, the two things were always intertwined, because I worked at an electronics shop (and before that, a camera shop). To date, I have come to the conclusion that I really enjoy working in organizations that are into some of that world. For example, I am on a bunch of Discord servers currently that may or may not be software based, but they’re related to fields where people think about physical constraints, or someone has to. There is a certain culture and mentality that is a part of that. It reminded me of my first robotics class, in which we were shown a video on the first class, and the professor, who worked on the Mars Rover Perseverance, said that he wanted us to know and understand that the things we built had impacts on people’s lives and could be detrimental (and even lead to death) if we didn’t care. There is a certain amount of pride in one’s work, integrity and commitment to craftsmanship that accompanies that line of thinking. It’s never just data in a repository. In film, you were very careful to prep the unit because this could be the one camera body or lens that you shipped to some camera person in the Amazon rainforest and if you forgot something, some tiny part, they couldn’t just “pick one up”, and in some cases it would mean they would waste hours, labour and unique opportunities they may never have access to ever again because of your negligence .
  • Systems thinking forces you to think from the ground up, and end to end, considering what could go wrong. It’s also thinking one step ahead.

Generative stuff

  • I’m taking a generative workshop for fun. We’ve done cellular automata, and made some L-trees. I think we’re doing a bit of reinforcement learning and generative stuff, as well. This is just a “I’m curious” class, and it’s more creatively focused and unaffiliated with my school (it’s in a space run by artists). Come to think of it, considering we learned about Merkle Trees and Reed-Solomon error correction and the Data Availability Problem in the context of scalability and sharding in my blockchain class, it’s pretty applicable :)

Next semester

  • Next semester, we’re supposed to be going back to campus, and I can’t say I’m thrilled. I’ve really enjoyed my time in quarantine; it feels like I was a good girl and someone decided to give me a treat by telling me I didn’t have to go anywhere :)
  • I had an incredible amount of green tea ice cream and matcha, enough to think that if I travel right now, the TSA might confuse me for an insect if asked to give a blood sample.
  • I’ll miss the hot meals, the power naps, and being able to take control of my time in ways you can’t really do on campus. Maybe there’s a way that I can plan so that next semester, I can limit my time on campus. I hate running into people, who then drag me all over the place “come to my office in an hour”, because it means I can’t control my time, and then I lose hours of my day that I won’t regain. So I want to reclaim that, but we’ll see.


  • I guess that’s it for now.
Written on March 20, 2021