wow ok what a smorgasbord of interviews. just got out of yet another interview i was vastly underprepared for
so far the technical-ish interviews i’ve had are
r&d engineer
machine learning engineer
data engineer
mechanical engineer
data engineer – sql and python questions over a virtual session. speed important. simple ones (1st round) machine learning engineer – transformers architectures and layers r&d engineer – mix of computer vision / image processing, simple physics, mechanical debugging questions mechanical engineering – high school physics, stress strain, mechanism design questions
going from like drawing the cross-sectional free body diagram to describing how to use isnull() on sql queries to find duplicate records is like
a lot
and i don’t get the chance to build on each interview since they’re so different in format and topic. i swear i can re-learn the stuff and am open to learning / re-learning but jeez what a way to feel like i don’t know anything in particular and i’ll be that worst-case coworker who doesn’t know what a for loop is
a common thing at least in hardware i’m realizing is the whole idea of going from first principles rather than trying to be a little more creative. the goal is to show you know the basics rather than that you are some shiny candidate
day 19: well, i have insomnia apparently so there’s that
alright i’m too lazy to type it like it’s all a big quest.
sailing / pfd
yesterday was what is likely my last sail of the season. feel really fortunate to have been on the water so much this year. probably 30-40 hours.
actually set up my pfd (personal flotation device), came with co2 cartridge that i never inserted. sailed so little after i got it that wasn’t worth it for years. it was literally just depress yellow lever thing on the bottom, slide in cartridge, press down hard against springs, and rotate. Helped a lot to look at the mechanism the cartridge slid into first.
bike
fixed my other mode of transport, my bike, which somehow i have continual issues with. i forgot i have hydraulic brakes so actually fixing the brakes should hopefully be adding some oil. on normal brakes, this feeling would be the pads wearing out or the cable somehow getting completely loose or something bizarre. though of concern is why there is so little oil. learned that my bike does have “oversized” brakes for the size of wheel it has, which is good given that this is 50lb-heavy ebike not bike.
typst
side note: typst is amazing! finally, a latex replacement. sorry overleaf
the equations! you don’t have to type \frac{a}{b}, just type a/b. bolding! no \textbf(hello) world, just *hello* world. the compile errors actually are concise and easy-to-understand.
you can even just use unicode directly in the editor, though I don’t show it below since i find it easier to search for “lambda” than λ.
makes me want to learn rust to work with these people…
meche roots
have a pure mechanical engineering interview coming up. feel a bit bizarre about that. i guess i can keep learning software engineering during my side projects?
reverse engineering
talked quite a bit to my friend about projects to do. (is it better to err on side of naming or not-naming people?). i mentioned how i wanted a portfolio project without feeling like i was working on a portfolio project. he brought up the good point of reverse engineering existing products as a way to learn good engineering ! and then try to improve on them.
then he gave me a shimano step ebike motor heh. my place is finally set up enough, with the workbench clear finally, to feel motivated to work on projects. just in time for my interview where hopefully i can explain how this ebike motor works.
i guess that is how i could have improved my coding skills also: reverse engineering existing good software. it feels different though. i don’t really have anyone to talk to about that sort of stuff except at a high level. i guess that was the nice part about miters and the meche ee sphere, there were just objects to stare at together on a bench.
thesis -> publication
have made no progress turning my thesis into a paper. hopefully that will change as sailing drops off (rip) and the lecture notes end (there’s only one month of them). anxious that i will be anxious once i stop having so much work to do.
lecture notes / keeping busy during funemployment
here is why i think the lecture note task has been a great complement to the job application process (which feels a bit like tossing my work into the wind):
on my own, my side projects are just quirky quick projects that i am not motivated to polish vs. move on to the next idea
not projects that scream professionalism, just quirkiness
but if the project is helpful to someone else, i am motivated to polish it
it’s immediately useful to others
vs. job apps where you can potentially throw infinite work in to no end. my friend said at this point she turns in grant applications and immediately forgets about them until she hears back, then just builds on it for the next application
there’s a deadline, which prevents me from just thinking about the best way to do it
there’s a time limit after which it stops being so useful
forcing me to focus on getting the core done (a minimum viable product) instead of polishing small sections to perfection
then i can go back and polish bits of it
(although it’s dozens of hours of volunteer work, and part of me does think, wow it’s an unfair advantage to (profs here) that there are people willing to just do this stuff for free lol, but the other part of me is like, we went over this, if you turn everything into a societal systems problem you will lose yourself as an object of concern, you decided to give yourself permission to just think about yourself and be happy instead)
the end, mostly because i am tired of staring at this screen.
found out earlier this week about two pirate’s hangouts*, one at hahvahd (harvard science data and tech fair) and one at seaport (roboboston)** . consulted the ol’ sailing logs (work history / resume). found some old treatises (6.7920 lectures recordings from last year) and creating some reports to send to the general assembly back home (lecture notes for the TAs and students based on those). then a bunch of scallywags (friends) decided to hold get-togethers (parties) too! well, the decks needed a good scrubbing (my place is a disaster) so now yours truly is having folks over as well.
(following is a ramble)
log for The Quest
in pursuit of one particular opening i’ve applied via website and linkedin, then also dug out a recruiter email and included my resume and cover letter, and also messaged another recruiter from the past on linkedin.
this recruiter then connected me to another company and i have an interview today for a software engineer position. they asked for a portfolio / open-source project contributions :melty face:.
on top of that i got a laundry list of things to refresh / learn and feel insecure (everyone else knows this), anxious (how long will it take to learn this), upset (why don’t i know this already), and frantic about (how am i supposed to do this for every type of position i’m applying for, should i study 2.001, chalkboard physics, learn transformers architecture, do labs, take an nlp class, what about making a deployed software project to learn system engineering tradeoffs between concurrency and latency, well i can talk about it from the hardware perspective of choosing cheap sensors vs fancy software, volume vs manufacturing method, i guess, but all i know really is derpy personal projects, and i’m not a fresh undergrad anymore,
and then it devolves into thinking about all the things i don’t know instead of the things i do. need to learn to not flounder and instead think deeply about the things i know and relate to things i have done and find fun vs. feel insecure and try to puff up directly related but shallow / derivative projects i guess?)
hitting the books
learning RL to “teach” (not really teach but in the spirit of it) has been fun and motivating. my ML knowledge really stalled out about 10 years ago (omg) when i last took an ML class. feels really good to gently touch math again, even if it is giving anxious flashbacks to my terrible RAI interview (that i sort of completely wasted, i mean the glass half full is i got free interview training about what questions those interview ask). some part of me still deeply believes that once my friends realize how little i know they’ll pity and think less of me.
but it was really neat to dig into every part of a slide. and use llm’s to keep me from getting stuck / stuck on notation and then lost. and wow the textbooks are actually really well written if i don’t understand anything in the slides. but dang no way i had time for this in undergrad.
decided to mimic mit class schedule instead of trying to cram in 10 lectures at once at 2x or something when anxiety strikes, just commit to being funemployed for a while, don’t think about being behind in life, “pace” my enthusiasm, and also ok if i don’t have prereqs, ok if i missed the first 5 lectures, ok if i slept through 2nd half of all the lectures, don’t keep going back to hear every word, but focus on the big picture. the real learning is in doing the psets and getting help on those
in general i think the idea of learning to teach something will get me to really dig in. so perhaps upcoming is a series of blog posts about transformers or who knows what. well-trod ground but good for me to practice at.
looking forward to feeling really confident in my knowledge of this material, enough to explain it to other people. forgot what that felt like. did i ever know it? i can feel that this would make interviews a lot better.
aside
enjoyed one lecture where the TA said (not for this class) “When I first took this class I failed it. Now I’m a TA. So yea, anything can happen” reminds me that my grades aren’t everything, that there is value in being able to complete projects, and even a friend who got a C in a class and went on to use it in a professional capacity daily.
bad ideas
trying to focus on keeping it lighthearted and fun, to manage my anxiety so i keep applying to jobs (though honestly the 3x/week to be able to cover living expenses gives me a lot of momentum there. this would have been so helpful in the past, to be able to apply to jobs :/ it’s honestly a stable paycheck for stuff i’m already trying to do vs. frantically searching for contract work), i came up with a bad (no, terrible) idea of making paperclips that say “best resume here”. (came from a comic where the reviewers put stickers to flag the best resumes).
anyway then i thought i should probably also put my name on it in case the clip fell off. then my roommate asked me if i want a job or for strangers to think i’m quirky. which was like a, sobering, need a job. i guess in the past my sheer enthusiasm has resulted in opportunities, but i’m past that point now and need to be more professional? then i also added dr. because idk it can be hard to find me online.
but then the end result makes me look like a prick …
onshape single color print — it works! paint tool in bambu slicer multi color print
in any case it was fun to learn onshape’s text tool (in its sketch interface) and also bambu lab’s painting interface (multicolor print).
pirate’s hangout
now hella late for the career fair. need to print at library 10 copies, so find 10 jobs and apply for them right now.
the ideal workflow would have been: find list of companies and positions apply online message a recruiter on linkedin or cold email (premium is expensive T_T) print out resume/cover and physically hand to recruiter (add some impressive 3d trinket i designed?)
really need to add a CLI flag for the resume styles, and modify the code so that it flips between a list of all the experiences. so i guess name all the items? and a config file for each company? that’s probably going a bit far. just on for software and one for meche i suppose. and maybe another file that is just more projects i have done…
