Tim Kanarsky

About Me

Hi! I'm a generalist software engineer, which means I'm mediocre at a lot of things at once :) I graduated from UCLA in 2023, and have been working at Tesla since then.

I'm interested in stuff like:

Robotics control, path planning, and vision algorithms.
Embedded software -- interactive art exhibits, home automation, IoT gadgets, etc.
Low-level software -- device drivers, high-performance compute,
latency and throughput optimization through knowledge of the underlying hardware.

More generally, I enjoy working on projects involving software, electrical, and mechanical design. I'm a big fan of Michael Reeves and Stuff Made Here on YouTube. I aspire to someday be on Shane's level of programming, CAD, and fabrication mastery :)

Friends

These are my friends' websites. Check them out!

Christian Aguilar, a fellow appreciator of the invisible infrastructure of the world, hiking, and beer. We studied together at UCLA and would hang out a lot after class, bonding over shared childhood experiences (ham radio, Boy Scouts, making DIY automatic sprinklers...)
Simon Safar, one of my coworkers at Tesla and modern philosopher who has already thought every single thought I have thought to be original.
Julie Cover, one of my former Makerspace coworkers who is particularly good at making computers work.
Matt Craig, my college roommate, who now lives in Austin with his wife and writes fast cpp at a quant firm.

Books I'm reading or have read

The Perfectionists - a book about how precision was bootstrapped and how it revolutionized manufacturing. Reading up on ball bearing tolerance categories made me wonder what the process was to get from "rolling logs to move big stones" to repeatable, characterized, commoditized thingies you could buy on Amazon for $5 a pack.
Bad Blood - about Elizabeth Holmes' Theranos saga. It's crazy how she got away with it for so long.
Daemon - a pretty realistic take on how a rogue non-human agent could take over the world and participate in the human economy, undetected, until it's too late.
To Say Nothing of the Dog - cozy novel about time travel to Victorian England and the hijinks that ensue.

Do things with me!

I have a ham radio license - my call sign is KK6EKQ. I haven't upgraded to general class yet (I've been meaning to do that.) I don't monitor VHF/UHF very often but let me know if you're on the air in the Bay Area -- let's QSL!

I like geocaching a lot. It's a good excuse to go outdoors and look incredibly suspicious while playing the IRL equivalent of a point and click mystery game. If you like it too, let's go caching on the weekend!

I walk the Stanford dish trail pretty often. It's by the office and the views are nice. Let me know if you have similar plans!

I absolutely adore corgis. If you have a corgi, and it likes to be pet, I want to pet it. The same goes with horses. I read online that a horse's nose is surprisingly soft. I want to pet a horse and feed it carrots but I don't know anyone with a horse. If you happen to be reading this and have a horse I could pet, I would love to do so.

Experience

Been writing code for a living, as you can see.

Robotics engineer at Tesla

I'm working on the Optimus humanoid robot. I've dipped my toes in basically every part of our robotics stack at this point.

Software engineer intern at Facebook/Meta, summer 2022

Did database refactoring and some fullstack work on the Posts Infra team.
Was pretty interesting seeing how huuuuuge-scale-software companies tick.
Wasn't really my cup of tea in the end, but I got to enjoy New York for a summer!

Flight Software intern at SpaceX, summer 2021

Designed, implemented, and tested a fault-tolerant protocol for configuring data flows on board the SpaceX Starship avionics network.

Avionics intern at SpaceX, summer 2020

Wrote device drivers and middleware for various automated test equipment (ATE) used in the avionics testing process.
Later helped optimize a telemetry validator algorithm using vectorized code and parallelism
Created a software framework for a new type of test procedure, combining ATE control, data validation, and technician-friendly anomaly reporting.

Projects

Pinecil rich presence integration for Discord

A Pinecil is a cheap USB-C powered soldering iron that for some reason has a bluetooth radio.
I wrote a little daemon in rust to poll the iron for its current state and expose that to Discord.
It was a silly challenge to learn how BLE works and to get better at understanding async rust. I still suck at the latter.

Tri-Valley Vaccine Radar

Wrote a Twitter bot that scrapes COVID vaccine appointment data from Rite Aid, CVS, and Safeway pharmacies in my area, and tweets about new openings.

Homebridge Alarm Listener

Homebridge is a platform for connecting third-party devices to Apple's HomeKit home automation framework.
Apple defines certain "device classes" that your IoT gadgets can appear as — switches, humidifiers, sensors, etc.
I wrote a Homebridge plugin that appears as a "smoke sensor" accessory. It uses the host computer's microphone to listen for smoke alarms going off in the area, and if it hears an alarm going off it triggers the virtual smoke sensor.
If the user has HomeKit Remote Access enabled, they'll be notified of their smoke alarm going off even if they're away from home.
They can also use the sensor value in automation scripts. For example, one can turn on the lights throughout the house if an alarm goes off.

Amador Valley Robotics software stuff

We designed, built, and tested an autonomous submersible, and competed in the annual RoboSub competition in San Diego.
Wrote a driver to interface a new velocity sensor with the sub's ATMega microcontroller over a serial link. It helped us navigate underwater much more precisely than before.
Created a web-based admin panel that allowed us to arm or disarm the sub, view feeds from cameras, monitor the internal state of the sub's software, and manually control the thrusters over a websocket. Saved a bunch time when debugging vision algorithms, as we could see exactly what the sub was seeing in real time.
Prototyped a series of object detection algorithms to recognize underwater obstacles such as colored buoys, floating gates, and plastic bins on the seafloor. I used the OpenCV library to develop routines that color-corrected the images, detected edges and other characteristic features, and clustered them into object bounding boxes.