A curated selection of public engineering work spanning AI agents, machine learning, computer vision, responsible AI, simulation, data science, and Web3.
These are public repositories selected from the GitHub profile. They are shown as engineering portfolio work, not represented as client engagements unless separately verified.
An autonomous AI system that generates, audits, self-corrects, and prepares smart contracts for EVM and Solana from plain-English requirements.
Time-series forecasting using multiple machine-learning models and LSTM-based sequence modeling with feature engineering.
Video and image wall segmentation using SAM2 with custom overlays for visualizing model output across frames and images.
A vision-transformer workflow for skin-cancer image classification, including fine-tuning and TensorFlow Lite conversion.
Gender classification on FairFace with a ResNet50 baseline and fairness metrics to evaluate model behavior across groups.
Dynamic programming and reinforcement-learning experiments on the Blackjack MDP using SARSA and Q-learning.
A consolidated X-15 flight-dynamics study with simulation, LQR control, literature review, and ARX system identification.
An interpretable Airbnb nightly-price model built with the SEMMA workflow, data cleaning, exploration, transformations, and regression assessment.
Preprocessing and loading workflows for KGQA datasets and knowledge graphs including CWQ, SimpleQuestions, WebQSP, Freebase, and Wikidata.
The terminal interaction is more than decoration. It mirrors the way we think about systems: clear inputs, visible state, explicit execution, and traceable outputs.
// intelligent delivery loop
const system = discover(businessContext)
.define(requirements)
.build(capabilities)
.validate(riskAndQuality)
.deploy(production)
.observe(outcomes);
// evolve only when evidence supports it
if (system.signal > noise) {
system.evolve();
}Use the portfolio as a conversation starter, then tell us the business outcome you need.