I am an engineer based in Boulder, Colorado with 3 years of iOS app development experience and a bachelors degree in computer and electrical engineering from CU Boulder. I have published 3 major apps to the app store, and have skills in computer vision, ML Ops, embedded systems, data science, and circuit design. I am also a Boy Scouts of America Eagle Scout.
My Remote Controlled Coil Gun uses electro magnets to accelerate a ferromagnetic projectile. Additionally, it is controlled remotely using an iPhone app to aim and fire the rail gun.
In my graduate level real-time embedded systems course, my team built a physical “Fruit Ninja” bot that smacks falling ping-pong balls mid-air. We framed it as a rate-monotonic system: two periodic solenoid tasks release balls (fruits) at independent rates, and a faster servo (sword) task intercepts the latest drop. The stack is C++ on a Raspberry Pi with mmap’d GPIO, pigpio PWM, and mutex-guarded shared state; 3D-printed chutes and MOSFET driver circuits handle the hardware. We validated schedulability with RM analysis and timing tests for consistent mid-air hits.
In my Mechatronics course at CU Boulder my team and I combined the classic Rock 'Em Sock 'Em board game with the robot shadow boxing seen in the movie Real Steel. In our modernized version of rock em sock em the robots mimic the movements of the players. When the player punches, the robot punches, when they dodge, the robot dodges and when the player moves forward or backward the robot follows. Then, we positioned two robots in front of each other to fight.
EIT Vision is a research project that verifies the feasibility of using radio frequency (RF) signals for Electrical Impedance Tomography (EIT) imaging, offering an alternative to the traditional methodology of using direct current (DC). EIT Vision generates a 2D heatmap that visualizes the location of a conductive object placed within the 10 inch circular platform. To accomplish this, it transmits and receives RF signals between six transceivers with antennas arranged evenly along the circumference of the platform. The system measures the changing power strengths due to signal reflections or blockage from the conductive object, enabling image reconstruction and proving the proof-of-concept. This was my capstone project for my undergraduate degree in Computer and Electrical Engineering at CU Boulder. My roles were team lead and system architect.
LLMs are limited to the knowledge built into their training data, and even then they tend to hallucinate. Connecting knowledge bases to LLMs allows you to give an LLM access to relevant information it otherwise wouldn't have, and improves the quality of it's responses. One way you can represent that knowledge to an LLM is with knowledge graphs. This projects lays out the basics of giving an LLM the ability to construct knowledge graphs during user interactions, and search a knowledge graph to inform its responses. Two key tools it uses include the OpenAI API, and the Neo4j knowledge graph database.
AI Resistor Scanner is an iOS app I developed to remove the tedious task of decoding electronic resistor values using color code charts. Instead, the user can take advantage of a chain of advanced computer vision models that do this for you. It makes use of a 3 stage pipeline: Detect, Segment, Classify. First it detects the resistor using oriented bounding boxes so it can crop and rotate the image around the resistor to minimize variations passed into the next stage of the pipeline. Stage 2, segment, uses image segmentation to extract each color band. Finally, the classification stage takes in each color band and outputs its color. It currently has 800+ Downloads!
Skills:
OpenCV
Convolutional neural networks
Transfer learning
YOLO Pretrain, ResNet, Pytorch, Tensorflow
Object Detection, Instance Segmentation, Image Classification
Fast API, ML Ops, Cloud Computing
Swift Frameworks: AVKit, CoreML, coremltools (python package for exporting pytorch models to coreml)
Set up a data collection pipeline from production application to cloud data base to be used in future model training (ETL - Extract, Transform, Load)
Concurrency - managing API calls in app
Multi threading - Running stages in parallel to optimize performance
Validated value proposition through customer discovery interviews
Created an ObjectiveC wrapper for OpenCV to use in Swift
Accountability is key when building healthy habits. I developed habit homies to be a tool for self development junkies to not only track and analyze their own habits, but also be able to track the habits of their homies. It also allows you to interact with your homies by nudging them, congradulating them, or expressing dissapointment in their performance. It also has social safety mechanisms like allowing you to report and block certain users.
Skills:
Networking
Firebase Firestore
User Authentication
Swift Data
Cloud Notifications
SwiftUI
Swift Concurrency
Managed synchronization between local and cloud databases
Optimized firebase reads and writes to minimize costs.
Received and implemented user feedback
Philosophers have spent ages contemplating art of living, and recording their thoughts in books. Sometimes these ideas have the power to change your life. However, getting exposed to these ideas, and consistently applying them can be challenging. Stoic Mentors connects Chat GPT to the user's personal journal, and classic stoic texts allowing it to expose the user to the most relevant stoic ideas, and assist them in applying those ideas to their life in a meaningful and practical way. This app was even featured on the educational app store.
Skills:
OpenAI API (Swift & Python)
Firebase Cloud Functions
Vectorized Databases (Pinecone)
LLM Research Augmented Generation (RAG) using vectorized databases
Prompt Engineering, LLM Tools, LLM Agents, Langchain
LLM Observer A
MVVM, Dependency Injection
Core Data, Swift Data, Cloud Kit
RevenueCat, StoreKit
Swift Natural Language
Publishing and Marketing an App