LOADING...

Hello, I'm
SEBASTIAN
MONZÓN.

Full-stack engineer, independent researcher, and musician exploring embodied human-computer interaction, multimodal generative AI models, and music technology - alongside my work developing distributed systems at Cisco Meraki.

All content on this website was coded by me
Githubsmonzon360@gmail.com
I always dreamt of having a personal assistant keep me up to date on everything that was happening in the world. That's why I designed an Ollama personal-assistant, a locally-stored Large Language Model (LLM) for New York Times articles that use realtime text-to-speech (TTS) to read out its response. The model refreshes articles daily and is capable of summarizing current events, or any other questions a user may have about recent or past news.
Go to project
I always dreamt of having a personal assistant keep me up to date on everything that was happening in the world. That's why I designed an Ollama personal-assistant, a locally-stored Large Language Model (LLM) for New York Times articles that use realtime text-to-speech (TTS) to read out its response. The model refreshes articles daily and is capable of summarizing current events, or any other questions a user may have about recent or past news.
A custom-built environment controller for mushroom cultivation, featuring an ESP32 microcontroller, a custom PCB, and OLED display. The controller is able to atomize water, connect to the internet, power UV lights, control PC fans, generate heat/cold with a Peltier solid-state cooling device, and had a selector knob for user-interaction. The project was built to create an optimal environment for growing mushrooms at home, and I designed a custom PCB in KiCAD to interface with all the components.
Go to project
A custom-built environment controller for mushroom cultivation, featuring an ESP32 microcontroller, a custom PCB, and OLED display. The controller is able to atomize water, connect to the internet, power UV lights, control PC fans, generate heat/cold with a Peltier solid-state cooling device, and had a selector knob for user-interaction. The project was built to create an optimal environment for growing mushrooms at home, and I designed a custom PCB in KiCAD to interface with all the components.
An ongoing solo research project that uses a variational auto-encoder to generate percussive audio from hand gestures. The model is based on RAVE, a state-of-the-art neural audio synthesis model, and is trained on a dataset of percussive hand movements and sounds from a conga drum (played by my dad) at a high frame rate to capture fine hand movements. The goal of this project is to be able to play any drum or percussive instrument with hand gestures and a camera.
Go to project
An ongoing solo research project that uses a variational auto-encoder to generate percussive audio from hand gestures. The model is based on RAVE, a state-of-the-art neural audio synthesis model, and is trained on a dataset of percussive hand movements and sounds from a conga drum (played by my dad) at a high frame rate to capture fine hand movements. The goal of this project is to be able to play any drum or percussive instrument with hand gestures and a camera.
Since it's nearly impossible to coordinate who's fed our cat Kevin, I soldered together a WiFi enabled smart button that allows us to schedule feedings and track when Kevin was last fed. The device retrieves local time from an NTP server and displays it on an OLED display. The device was built with an ESP8266 microcontroller, an SSD1306 OLED display, and a button.
Go to project
Since it's nearly impossible to coordinate who's fed our cat Kevin, I soldered together a WiFi enabled smart button that allows us to schedule feedings and track when Kevin was last fed. The device retrieves local time from an NTP server and displays it on an OLED display. The device was built with an ESP8266 microcontroller, an SSD1306 OLED display, and a button.
In collaboration with researchers at the MIT Dept. of MechE and Media Lab, I developed a Stable Diffusion model and pipeline (based on Riffusion) to inpaint music spectrograms to improve hydrogel air-water extraction by increasing desirable frequencies in music. This project is ongoing and I am currently experimenting with CLIP soft-tokens to improve the generative guidance of the model.
Go to project
In collaboration with researchers at the MIT Dept. of MechE and Media Lab, I developed a Stable Diffusion model and pipeline (based on Riffusion) to inpaint music spectrograms to improve hydrogel air-water extraction by increasing desirable frequencies in music. This project is ongoing and I am currently experimenting with CLIP soft-tokens to improve the generative guidance of the model.
A quadcopter built with a Raspberry Pi, featuring a camera, GPS, and Bluetooth/WiFi control. I designed this project to learn about quadcopter dynamics and attempt to code my own flight controller using PID. Unfortunately, A Raspberry Pi has many drawbacks to be used as a flight controller, but I taught myself PID control and quadcopter dynamics to try it the HARD way.
Go to project
A quadcopter built with a Raspberry Pi, featuring a camera, GPS, and Bluetooth/WiFi control. I designed this project to learn about quadcopter dynamics and attempt to code my own flight controller using PID. Unfortunately, A Raspberry Pi has many drawbacks to be used as a flight controller, but I taught myself PID control and quadcopter dynamics to try it the HARD way.
As a music producer, I struggled with engineering the right sound textures and timbres I was looking for. That's what inspired me to research Generative Adverserial Networks for timbre synthesis. Though I didn't have the time or resources during undergrad, I developed a multiclass-classification model for instrument identification. The idea was to use a GAN to synthesize audio with specific timbre descriptors from text (i.e. "soothing piano with long release and pleasant overtones"). In theory, an AI model that did this would be incredibly useful for sound designers/music producers to come up with new sounds or use words for sound synthesis. This was my undergraduate honors thesis at the UMass Amherst Commonwealth Honors College.
Go to project
As a music producer, I struggled with engineering the right sound textures and timbres I was looking for. That's what inspired me to research Generative Adverserial Networks for timbre synthesis. Though I didn't have the time or resources during undergrad, I developed a multiclass-classification model for instrument identification. The idea was to use a GAN to synthesize audio with specific timbre descriptors from text (i.e. "soothing piano with long release and pleasant overtones"). In theory, an AI model that did this would be incredibly useful for sound designers/music producers to come up with new sounds or use words for sound synthesis. This was my undergraduate honors thesis at the UMass Amherst Commonwealth Honors College.
A physics simulator built with Python and Pygame, featuring a 2D physics engine that simulates gravitational bodies and instantaneous velocities in a 2D space. The simulator allows users to create and manipulate objects in a 2D space, and observe their interactions with each other. I built this project to learn about PyGame and test my knowledge of basic gravitational physics.
Go to project
A physics simulator built with Python and Pygame, featuring a 2D physics engine that simulates gravitational bodies and instantaneous velocities in a 2D space. The simulator allows users to create and manipulate objects in a 2D space, and observe their interactions with each other. I built this project to learn about PyGame and test my knowledge of basic gravitational physics.
Originally I wanted a way to "clean my room remotely", so I designed this 3-wheel claw robot to pick up objects with a PS4 controller, featuring a camera, an ultrasonic sensor, and dynamic claw with 2 degrees of freedom. The robot is capable of using OpenCV for object detection and grabbing objects. I partially designed this robot as an excuse to experiment with computer vision and object detection.
Go to project
Originally I wanted a way to "clean my room remotely", so I designed this 3-wheel claw robot to pick up objects with a PS4 controller, featuring a camera, an ultrasonic sensor, and dynamic claw with 2 degrees of freedom. The robot is capable of using OpenCV for object detection and grabbing objects. I partially designed this robot as an excuse to experiment with computer vision and object detection.
A C++ implementation of Conway's Game of Life, a cellular automaton that simulates the evolution of a grid of cells based on simple rules. The project was built to expirement with cellular automata and the elegant patterns that arise from simple rules. I color-graded cells depending on which rule they were following.
Go to project
A C++ implementation of Conway's Game of Life, a cellular automaton that simulates the evolution of a grid of cells based on simple rules. The project was built to expirement with cellular automata and the elegant patterns that arise from simple rules. I color-graded cells depending on which rule they were following.
I always dreamt of having a personal assistant keep me up to date on everything that was happening in the world. That's why I designed an Ollama personal-assistant, a locally-stored Large Language Model (LLM) for New York Times articles that use realtime text-to-speech (TTS) to read out its response. The model refreshes articles daily and is capable of summarizing current events, or any other questions a user may have about recent or past news.
Go to project
I always dreamt of having a personal assistant keep me up to date on everything that was happening in the world. That's why I designed an Ollama personal-assistant, a locally-stored Large Language Model (LLM) for New York Times articles that use realtime text-to-speech (TTS) to read out its response. The model refreshes articles daily and is capable of summarizing current events, or any other questions a user may have about recent or past news.
A custom-built environment controller for mushroom cultivation, featuring an ESP32 microcontroller, a custom PCB, and OLED display. The controller is able to atomize water, connect to the internet, power UV lights, control PC fans, generate heat/cold with a Peltier solid-state cooling device, and had a selector knob for user-interaction. The project was built to create an optimal environment for growing mushrooms at home, and I designed a custom PCB in KiCAD to interface with all the components.
Go to project
A custom-built environment controller for mushroom cultivation, featuring an ESP32 microcontroller, a custom PCB, and OLED display. The controller is able to atomize water, connect to the internet, power UV lights, control PC fans, generate heat/cold with a Peltier solid-state cooling device, and had a selector knob for user-interaction. The project was built to create an optimal environment for growing mushrooms at home, and I designed a custom PCB in KiCAD to interface with all the components.
An ongoing solo research project that uses a variational auto-encoder to generate percussive audio from hand gestures. The model is based on RAVE, a state-of-the-art neural audio synthesis model, and is trained on a dataset of percussive hand movements and sounds from a conga drum (played by my dad) at a high frame rate to capture fine hand movements. The goal of this project is to be able to play any drum or percussive instrument with hand gestures and a camera.
Go to project
An ongoing solo research project that uses a variational auto-encoder to generate percussive audio from hand gestures. The model is based on RAVE, a state-of-the-art neural audio synthesis model, and is trained on a dataset of percussive hand movements and sounds from a conga drum (played by my dad) at a high frame rate to capture fine hand movements. The goal of this project is to be able to play any drum or percussive instrument with hand gestures and a camera.
Since it's nearly impossible to coordinate who's fed our cat Kevin, I soldered together a WiFi enabled smart button that allows us to schedule feedings and track when Kevin was last fed. The device retrieves local time from an NTP server and displays it on an OLED display. The device was built with an ESP8266 microcontroller, an SSD1306 OLED display, and a button.
Go to project
Since it's nearly impossible to coordinate who's fed our cat Kevin, I soldered together a WiFi enabled smart button that allows us to schedule feedings and track when Kevin was last fed. The device retrieves local time from an NTP server and displays it on an OLED display. The device was built with an ESP8266 microcontroller, an SSD1306 OLED display, and a button.
In collaboration with researchers at the MIT Dept. of MechE and Media Lab, I developed a Stable Diffusion model and pipeline (based on Riffusion) to inpaint music spectrograms to improve hydrogel air-water extraction by increasing desirable frequencies in music. This project is ongoing and I am currently experimenting with CLIP soft-tokens to improve the generative guidance of the model.
Go to project
In collaboration with researchers at the MIT Dept. of MechE and Media Lab, I developed a Stable Diffusion model and pipeline (based on Riffusion) to inpaint music spectrograms to improve hydrogel air-water extraction by increasing desirable frequencies in music. This project is ongoing and I am currently experimenting with CLIP soft-tokens to improve the generative guidance of the model.
A quadcopter built with a Raspberry Pi, featuring a camera, GPS, and Bluetooth/WiFi control. I designed this project to learn about quadcopter dynamics and attempt to code my own flight controller using PID. Unfortunately, A Raspberry Pi has many drawbacks to be used as a flight controller, but I taught myself PID control and quadcopter dynamics to try it the HARD way.
Go to project
A quadcopter built with a Raspberry Pi, featuring a camera, GPS, and Bluetooth/WiFi control. I designed this project to learn about quadcopter dynamics and attempt to code my own flight controller using PID. Unfortunately, A Raspberry Pi has many drawbacks to be used as a flight controller, but I taught myself PID control and quadcopter dynamics to try it the HARD way.
As a music producer, I struggled with engineering the right sound textures and timbres I was looking for. That's what inspired me to research Generative Adverserial Networks for timbre synthesis. Though I didn't have the time or resources during undergrad, I developed a multiclass-classification model for instrument identification. The idea was to use a GAN to synthesize audio with specific timbre descriptors from text (i.e. "soothing piano with long release and pleasant overtones"). In theory, an AI model that did this would be incredibly useful for sound designers/music producers to come up with new sounds or use words for sound synthesis. This was my undergraduate honors thesis at the UMass Amherst Commonwealth Honors College.
Go to project
As a music producer, I struggled with engineering the right sound textures and timbres I was looking for. That's what inspired me to research Generative Adverserial Networks for timbre synthesis. Though I didn't have the time or resources during undergrad, I developed a multiclass-classification model for instrument identification. The idea was to use a GAN to synthesize audio with specific timbre descriptors from text (i.e. "soothing piano with long release and pleasant overtones"). In theory, an AI model that did this would be incredibly useful for sound designers/music producers to come up with new sounds or use words for sound synthesis. This was my undergraduate honors thesis at the UMass Amherst Commonwealth Honors College.
A physics simulator built with Python and Pygame, featuring a 2D physics engine that simulates gravitational bodies and instantaneous velocities in a 2D space. The simulator allows users to create and manipulate objects in a 2D space, and observe their interactions with each other. I built this project to learn about PyGame and test my knowledge of basic gravitational physics.
Go to project
A physics simulator built with Python and Pygame, featuring a 2D physics engine that simulates gravitational bodies and instantaneous velocities in a 2D space. The simulator allows users to create and manipulate objects in a 2D space, and observe their interactions with each other. I built this project to learn about PyGame and test my knowledge of basic gravitational physics.
Originally I wanted a way to "clean my room remotely", so I designed this 3-wheel claw robot to pick up objects with a PS4 controller, featuring a camera, an ultrasonic sensor, and dynamic claw with 2 degrees of freedom. The robot is capable of using OpenCV for object detection and grabbing objects. I partially designed this robot as an excuse to experiment with computer vision and object detection.
Go to project
Originally I wanted a way to "clean my room remotely", so I designed this 3-wheel claw robot to pick up objects with a PS4 controller, featuring a camera, an ultrasonic sensor, and dynamic claw with 2 degrees of freedom. The robot is capable of using OpenCV for object detection and grabbing objects. I partially designed this robot as an excuse to experiment with computer vision and object detection.
A C++ implementation of Conway's Game of Life, a cellular automaton that simulates the evolution of a grid of cells based on simple rules. The project was built to expirement with cellular automata and the elegant patterns that arise from simple rules. I color-graded cells depending on which rule they were following.
Go to project
A C++ implementation of Conway's Game of Life, a cellular automaton that simulates the evolution of a grid of cells based on simple rules. The project was built to expirement with cellular automata and the elegant patterns that arise from simple rules. I color-graded cells depending on which rule they were following.
P
A lifelong learner ready to make a difference...

Designer

3d

4websites

Developer

3d

4Y.O.E.

Graduate

Edu

'23UMass Amherst

Engineer

3d

18+Projects

Musician

3d

100+Songs

SKILLS & TECHNOLOGIES

I'm constantly looking to solve problems with what I know, and searching for opportunities to learn what I don't.

  • ✔ HTML, CSS, & JavaScript
  • ✔ Python
  • ✔ Ruby on Rails
  • ✔ C/C++
  • ✔ Node/React JS
  • ✔ KiCAD
  • ✔ Fusion 360
  • ✔ Ableton Live
  • ✔ Adobe Suite
  • ✔ Matlab
^ Coded by me on GitHub and CodePen.
CURRICULUM VITAE

Whether it's building innovative software, engineering electrical systems, or exploring the intersection between AI and music, I strive to push my creative boundaries through experimentation and learning from failures.

Below you will find an incomplete list of the things I've made. The list is always growing, as I continue to take on harder challenges aligned with my passion for hands-on engineering, science, and music. I hope you enjoy.

Sincerely,
Sebastian Monzón

Download CV
Research

Research

See my research on AI, music technology, and human-computer interaction.

View Research
Engineering Projects

Engineering Projects

See my robotics, 3D design, electronics, and IoT projects.

View Projects
Software Projects

Software Projects

See my software projects which incorporate AI, simulations, mobile apps, commercial websites, and games.

View Projects
Music

Music

Listen to my latest tracks and remixes.

Listen Now

PROJECTS

RESEARCH IN-PROGRESS

MAGE: Motion-to-Audio Generative autoEncoder

Hand landmarks are extracted via MediaPipe and augmented with normal vectors to capture hand orientation.

Music-Spectrogram Inpainting

The framework I developed to with my Riffusion inpainting model. The model takes in a spectrogram with a masked region, and generates new audio in the masked region while preserving the unmasked regions.

The framework I developed to with my Riffusion inpainting model. The model takes in a spectrogram with a masked region, and generates new audio in the masked region while preserving the unmasked regions.

Honors Thesis

A confusion matrix showing the performance of the classifier model on a test set. The model was able to achieve an accuracy of 76% on the test set, with confused classes being psychoacoustically similar. e.g. (bass vs. guitar) or (reed vs. brass.)

A confusion matrix showing the performance of the classifier model on a test set. The model was able to achieve an accuracy of 76% on the test set, with confused classes being psychoacoustically similar. e.g. (bass vs. guitar) or (reed vs. brass.)

SOFTWARE

NYT Large Language Model (LLM)

Here is a video of the NYT LLM in action, answering my question "What is happening in the tech sector today?".

Beamshyft - Lower construction costs for developers

The website for a service that delivers interior construction materials to developers. This is a startup I'm building with a friend, and we are currently working on building the service.

The website for a service that delivers interior construction materials to developers. This is a startup I'm building with a friend, and we are currently working on building the service.

Curdle - Wordle for Cheeses

A web app that allows users to play a wordle-style game with 5-letter cheese names. (There are more than you think)

A web app that allows users to play a wordle-style game with 5-letter cheese names. (There are more than you think)

Newtonian Physics Simulator

A physics simulator built with Python and Pygame, featuring a 2D physics engine that simulates gravitational bodies and instantaneous velocities in a 2D space.

Conway's Game of Life, C++

A C++ implementation of Conway's Game of Life, a cellular automaton that simulates the evolution of a grid of cells based on simple rules.

Blockbreaker

A modern blockbreaker game built with JavaScript and an HTML5 canvas.

VibeQ - Spotify group DJ app

"Still waiting for your favorite song to play? Bring aux to the bourgeoisie with VibeQ!"

"Still waiting for your favorite song to play? Bring aux to the bourgeoisie with VibeQ!"

MyHS - Highscool student portal

A web app that syncs highschool grades, homework, and schedule all in one place (logo designed by me). The app was writen in Swift while I attended Melrose Public highschool.

A web app that syncs highschool grades, homework, and schedule all in one place (logo designed by me). The app was writen in Swift while I attended Melrose Public highschool.

ENGINEERING

FeedKevin! - Pet Food Timer

A gif of the pet food timer in action, using an ESP8266, SSD1306 OLED display, and a button. The schedule is set programmatically, and the button is used to move between interfaces when the timer is running.

Raspberry Pi Robot

The robot's top view, showing the camera and ultrasonic sensor. The camera was used for object detection and the ultrasonic sensor was used to determine distance to objects.

The robot's top view, showing the camera and ultrasonic sensor. The camera was used for object detection and the ultrasonic sensor was used to determine distance to objects.

Raspberry Pi Bluetooth/WiFi Quadcopter

Parts of the quadcopter, including a Raspberry Pi, camera, GPS module, ESC management, and IMU. The drone was designed to be controlled via Bluetooth or WiFi, and I used a Raspberry Pi for the flight controller.

Parts of the quadcopter, including a Raspberry Pi, camera, GPS module, ESC management, and IMU. The drone was designed to be controlled via Bluetooth or WiFi, and I used a Raspberry Pi for the flight controller.

Mushroom Environment Controller

A custom-built environment controller for mushroom cultivation, featuring an ESP32 microcontroller, and a custom PCB with several interfaces.

A custom-built environment controller for mushroom cultivation, featuring an ESP32 microcontroller, and a custom PCB with several interfaces.

3D Printer Cooling System

A custom-built cooling fan, mount, and nozzle system for a 3D printer. I built the original 3D Printer from a kit, and I designed this cooling system to improve the printer's capabilities and print-quality.

A custom-built cooling fan, mount, and nozzle system for a 3D printer. I built the original 3D Printer from a kit, and I designed this cooling system to improve the printer's capabilities and print-quality.

Anybody Home?

An ESP8266 that detects when someone is home by scanning for their phone's MAC address.

An ESP8266 that detects when someone is home by scanning for their phone's MAC address.

Breaker Panel Monitor

A custom-built breaker panel monitor that detects when a circuit is tripped and sends a notification to your phone.

A custom-built breaker panel monitor that detects when a circuit is tripped and sends a notification to your phone.

MUSIC

Contact

Contact

^ A Gibson les Paul guitar, painted by me :)

smonzon360@gmail.com(781) 530-8016LinkedIn