Luke Atkins

HackHavard W 🏆🥾 - Health and Fitness Track, biometric trail grading

Competed at HackHarvard 2023 in a team of four awesome engineers and designers! We built GreenTrail, an app that grades hiking trails based on metrics from users' biometric data while performing the trail. GreenTrail allows anyone to view trails and accurately assess their physical preparedness. By aggregating past users' HR, BP, perspiration, and other data while on the trail from their wearables, we create distributions from which to compare others. We used TerraAPI, connecting to any wearable device to collect health and fitness data. The app, built in SwiftUI, uses Firebase for user and image storage. For design, Figma was utilized.

Brewed with ❤️ using Caffine, SwiftUI, Firsebase, Figma, and Maps.

CashMap 💵📍 - IOS app for location based spending awareness

CashMap is an IOS app that promotes daily budget mindfulness in a geographical context. Users record their purchases, which are then displayed on a visual map of their local area. CashMap enhances spending awareness by sending push notifications when users approach their spending hotspots and provides widgets for at-a-glance spending metrics. Whether you aim to kick buying coffee out everyday or prioritize supporting local businesses, CashMap reminds you.

Baked with ❤️ using UI Kit, Swift UI, Widget Kit, Map Kit, Core Location, and Core Data.

Spotify Artist Niche Analysis 🎶📊 - SpotiPy API and regression

Analyze musicians' specialized features using Spotify API data. By regression on popular songs' abstract features, we uncover each artist's niche. Factors like danceability, energy, instrumentalness, etc., contribute to the analysis. For the aspiring artist or establsihed band, this serves as a reference point to connect with their fan base. For the music enjoyer, it enhances their musical literacy, enabling them to communicate and explore preferred abstract musical features within their own playlists and beyond.

Basted with ❤️ using Python, R, SpotiPy, NumPy, and Rpart.

NumPy Only Neural Network 🧠🍷 - Wine Dataset

Wine region classification achieved by building a neural network from scratch using a Multi-Layer Perceptron algorithm. The implementation is based on 'Machine Learning: An Algorithmic Perspective' by Stephen Marsland, Chapman & Hall. The entire model is implemented in NumPy without the use of pandas or any other machine learning libraries. Although this model does not replace the expertise of a sommelier, it achieves an impressive accuracy of 96% in 5-fold cross-validation with just one hidden layer.

Battered with ❤️ using NumPy

Knowledge Map 🤔🗺️ - SpartaHack 8 hackathon submission

Knowledge connects ideas for you by geographically displaying interactable nodes related to a query. Using the OpenAI API, KnowledgeMap is able to recurse and expand on topics related to your search. KnowledgeMap enhances your website with its meticulous approach to learning guidance. The map design encourages learning with a literal explosion of ideas as seen here. Valuable for self learning and research, KnowledgeMap is much like if Wikipedia and Google had a baby!

BBQ'd with ❤️ using Flask, Python, CSS and JS

Arduino Pinhole Touch 🤖🔌- Potentiometer, accelerometer, light sensor

The next iteration of "Pinhole". I aimed to add a physical interface that plays with the relationship between action and reaction. This idea, stepped toward using an Arduino, allows participants to form their own models of how the piece reacts to their actions. Exemplifying the concepts of constructivisim and information, the physical board offers a canvas for participants to learn about the piece through play. A slide deck about the work can be found here.

Pinhole 🌘🌊- Live Exhibited Work, light and motion interactive

"Pinhole” is meant to mimic the pinhole phenomena during an eclipse where the pinholes of light that poke through a tree's canopy cast an image of the sun onto the ground bellow. Through visual distortion, the viewers form is instantiated multiple times and presented back to them just like the crescent shapes act as copies of the eclipse. In this sense, the viewer is the moon. As an added interactivity, “Pinhole” reflects the change of light to dark during the eclipse; blocking all visible light from the performative camera changes the piece into a flurry of lighting bugs. While with full light, the view is of the users self and a dandelion blowing in the wind loosing its seeds.

Remix and Rememory ⏳🌌- Live Audio Reactive, for live music sets

"Remix and Rememory" consists of three multimedia scenes. Scene 1 depicts a black swirling galaxy ball expanding to engulf the entire frame, inspired by old video game aesthetics. Scene 2 shows hands partially submerged in remixed lava, with visible watermark stains representing appropriation in remix culture. The frame diminishes as noise increases. Scene 3 abandons any semblance of a frame, allowing incoherent noise to overtake coherent shapes and objects overlaid on the visuals. The scenes explore chaos, manipulation, and acceptance within online remix culture through stylistic choices like synchronized audio-visuals and fading effects on the hands. The piece contemplates complexities of digital creativity.