Common nixos configurations for all devices defined in a declarative manner

Led a backend team of 5 and designed an open-source application where fans can curate ideas about exhibits and artifacts to be held in Comic-Con Museum, San Diego with Express.js and AWS Lambdas

Implemented a multiprocessing Search Engine, Snippet Generator and Web Crawler with a MAP greater than Lucene combining several retrieval models such as BM25, Smoothed Query Likelihood and TF-IDF

Local Ollama Setup with Home Assistant with Matter Workflow

Developed an autonomous StarCraft bot using Convolutional Neural Networks (CNN) powered by TensorFlow and OpenCV. The AI agent processes real-time game frames to detect units, buildings, and resources, making strategic decisions with 85% accuracy in basic combat scenarios. Implemented custom computer vision algorithms for map awareness and unit control, achieving a competitive performance against medium-difficulty AI opponents in micro-battles.

Built four Music Transformers for each track inspired by Google Magenta and applied Global Multi Headed Attention allowing all decoders to attend over all encoders from all Transformers modifying the traditional architecture of a Transformer Network

Trained Matrix Factorization models on Gutenberg to automatically extract hypernyms from patterns with NLTK and Gensim

Analyzed several models in Tensorflow for generating paragraph representations using VAE and Unidirectional LSTM Seq2Seq with/without Attention on Feature rich word embeddings retrieved from the Gutenberg Corpus

Intsalled a Home Server on a N100 mini PC and a Raspberry Pi cluster using Proxmox, Pfsense, Home Assistant, Open Media Vault, Tailscale and Cloudflare DNS

Constructed and Soldered a Corne Keyboard from scratch with custom ZMK firmware with nice nano and nice view devices

A self driving robot with basic inverse kinematics and ollama integration with raspberry pi and tachyeon

Experimented with ROS2 suite for simulation training for ARM based robots

Analyzed effectiveness of Transfer Learning on Robot Arm Simulation Training using PyTorch and Open AI Gym on various algorithms such as Q Learning, DQN, Policy Gradient and Markov Chain Monte Carlo achieving significant advancements.

Created a Second Brain workflow using AI with Obsidian and kick ass open source plugins