I'm Jared Di Carlo (sometimes spelled DiCarlo). My interests are robotics, computers, computer graphics, reverse engineering, programming languages, and video game emulation.
I was at MIT from 2015 to 2020, studying Computer Science and Electrical Engineering. During that time, I worked with Professor Sangbae Kim in the MIT Biomimetic Robotics Lab. I helped with the electronics and software for the Cheetah 3 robot. I wanted to make robots run and jump, so I studied control systems and convex optimization and developed a Model Predictive Controller that you can see in the video below:
After that, I helped Ben Katz on his Mini Cheetah robot. I worked on the software and control software, moving away from MATLAB to a pure C++ controller. I got my first taste of nonlinear optimization in my final project for the class Underactuated Robotics - optimizing a backflip trajectory for Mini Cheetah.
From January 2020 to September 2025 I worked at Boston Dynamics. I worked on a ton of different things there, including model predictive control (MPC), writing really fast linear algebra for the MPC optimizer, motor control, firmware, low-level actuator and joint control, general robot infrastructure code, and so much robot debugging. The most memorable projects are the dance video, where I worked on the tracking controller and the All New Atlas, where I worked on electric motor control and the low-level controllers for the fancy back and ankle mechanisms.
Since September 2025, I work at Physical Intelligence.
I have no relation to Jared DiCarlo on Twitter/X.
I do a lot of programming projects. Hopefully I'll find the time to document them in detail on this blog. One of my favorite is OpenGOAL, which is a port of the PlayStation2 video game Jak and Daxter to PC. The developers of this game were somewhat crazy and actually wrote the game entirely in their own LISP-based programming language called GOAL. I built reverse engineering tools for this language, including a decompiler and a compiler. With the help of about 5 contributors, we decompiled millions of lines of GOAL code, ported platform-specific code to PC, then rebuilt the games as native x86 executables. Here's a sample of what GOAL code looks like:
(defstate plat-path-active (plat)
:virtual #t
:event plat-event
:exit (behavior ()
(sound-stop (-> self sound-id))
)
:trans (behavior ()
(set! (-> self path-pos) (get-norm! (-> self sync) 0))
(get-point-at-percent-along-path! (-> self path) (-> self basetrans) (-> self path-pos) 'interp)
(if (< (vector-vector-distance (-> self root trans) (ear-trans 0)) 81920.0)
(sound-play "eco-plat-hover" :id (-> self sound-id) :position (-> self root trans))
)
(plat-trans)
)
:code (behavior ()
(until #f
(ja-no-eval :group! (ja-group) :num! (seek!) :frame-num 0.0)
(until (ja-done? 0)
(suspend)
(ja :num! (seek!))
)
)
#f
)
:post plat-post
)
I also like embedded projects and video game emulation.


Outside of programming, I like video games, playing piano, and running.
This wouldn't be a programming blog without a section about my computer.
My main computer is named VCR3, as it's the third upgrade of my once VCR-shaped computer.
I use the i3 window manager. For text editing, I use a mix of vim, Sublime Text, and Jetbrains CLion. I've recently started using VSCode and Zed.
As of October 2025, my desktop looks something like this.

I also have a M2 Macbook Air for web browsing.