Hi, I am Max. I like learning how computers work.
I am currently a 4th year computer engineering student.
- Designed and built a webhook proxy service that is now used to forward thousands of payloads per day by over 15 production projects. The service simplifies existing webhook issuance process and improves the security stance for projects that have to use webhooks.
- Worked on adding new AWS resources to Geoengineer (infrastructure as code DSL based on Terraform).
- Helped organize and create cryptography challenges for Capture the Coin CTF at DEF CON 27.
- Worked on a hotel concierge virtual assistant that is capable of understanding and acting on hundreds of intents. Focused on conversation modelling and developing admin webapp.
- Implemented several conversation modelling techniques for an open-source chatbot utility library.
- Built a metadata search tool. Compared to the existing solution, query response time was improved by over 150% and ETL processing time by over 300%. Wrote ETL scripts to load data from new sources.
- Researched and evaluated several virtual assistant solutions. Developed a help desk virtual assistant proof of concept.
- Automated manual email distribution and account administration processes.
- Created a tool for tracking metrics of an analytics platform. Used Selenium to capture and track latency as experienced by users.
AI Doom player that was trained to interact with its environment using only raw visual information. This is my first time playing with reinforcement learning and trying out Tensorflow Eager Execution.
“DocuSign on a blockchain”. BlocuSign is a eSignature solution built on Ethereum; it also uses IPFS and BloomID decentralized identity attestation service. ETHSanFrancisco hackathon Bloom API prize winner.
Neon is an endless runner video game with a memorable visual design (mainly inspired by Tron). Game engine is written in Three.js. Myo armband is used to control all in-game actions. McHacks 2018 “Best Retro Hack” winner.
A blogpost describing my attempt at solving a wargame which required submitting CAPTCHA in a very short amount of time (impossible to do manually). Solving this challenge required me to collect and label seed training data, generate augmented training data, create and train a convolutional neural network, and write a sliding window algorithm.
Linear regression model with lasso penalty for House Prices: Advanced Regression Techniques Kaggle competition. This is my very first attempt at applying statistics/ML.