Max Wolff

dappTX (in progress)

code

Visualizes transaction volume of biggest ethereum smart contracts. Pulls data from Infura API, chron job populates a Postgres DB, Node/Express web app exposes the data with graphs on the front end.

• Supported as an open-source project by Gitcoin.co

PAD Piper

code

Winner of the Vasctrac Data Challenge at Health ++ 2017. We worked with a vascular health research group focusing on Peripheral Arterial Disease to interpret raw sensor data from a 6 minute walking test (accelerometer data from iphone, no gyroscope).

• Used DSP methods to create an 92% accurate step counter (iPhone is 94%)
• Built predictors for Ankle Brachial Index and distance walked

Heap Allocator

assignment

Pair prorammed a heap allocator in C for “Computer Organizations & Systems” at Stanford. Implemented malloc, realloc, and free functionality. Our allocator assigned memory addresses individually, optimizing for memory utilization, througput, and correctness.

• Used a double linked list of freed blocks embedded in the heap.
• Achieved 70% of the standard C library function’s utilization rate.

Roots

design process | download

Built a mobile app to promote community engagement on React.JS in an HCI class at Stanford. We started by interviewing our intended user base: civic-oriented community builders.

• We gathered feedback early and often, using an experience prototype, wireframes, interactive proto.io prototype
• Final product was a frontend app in react mobile via exponent

Lipreader

code | class

Machine learning project for our final project in AI Principles at Stanford. We were drawn to build a lipreader because of the complexity and direct use-cases for assisting people with hearing impairments.

• Takes in a short clip of a subject speaking one phoneme and produces an array containing the estimated probabilities that that clip corresponds to any particular phoneme
• We also built an NLP model, so that the phoneme classifier could be used to move from audio-free video to a full reconstruction of a spoken sentence.
• Implemented using Python, and Keras, and borrowing ideas for spatio-temporal convolutional neural network from Lipnet.

Tunnel

code | class

Audio-visual game for Music, Computing, and Design class with Ge Wang. Light up the tunnel by colliding with the spheres that come your way.

• Pick an audio track to play with. Used Aubio to parse music for drum onset so I could sync game elements with it.

Other Projects

Way Less Effort

videos: 1|2

Music collective to throw shows, make clothes, and created content for artists.

• Awarded two grants to host speakers on campus including Lyor Cohen (founder of 300, former CEO of Warner Music), Andre Power (founder of Soulection)
• Featured artists like Mikos Da Gawd, Dripmob and Ritchrd
• Executed marketing campaigns to sell Sold 350+ tickets to warehouse show with no budget