C. Kinkade Darling
Salt Lake City, Utah
6+ years at the intersection of technology and finance, shipping production ML systems end-to-end, from messy raw data to models running in prod. Comfortable going deep as an IC or zooming out to drive cross-functional work. I like hard problems, clean pipelines, and decisions backed by actual evidence.
Projects
Machine Learning · Finance
Crypto ML Forecasting
Two live forecast modes for BTC, ETH, and LTC: a 7-day-ahead point estimate and an autoregressive multi-step rollout up to 30 days. Both return 90% split-conformal prediction intervals. Built with feature engineering, two-pass importance pruning, chronological splitting, and three model classes evaluated on RMSE, R², and directional accuracy.
View project →Machine Learning · FinTech
Personalized Equity Discovery
A hybrid recommendation engine for retail equity discovery, combining Implicit ALS matrix factorization, Bayesian Personalized Ranking, and content-based filtering over GICS sector features. Five investor cohorts demonstrate cold-start handling, popularity-bias auditing, and intra-list diversity. All data is fully synthetic, calibrated to retail brokerage behavioral statistics.
View project →Deep Learning · Algorithmic Trading
Gordon — Autonomous Trading System
Paper-trading system managing a $100K simulated portfolio across 55 US equities. “Gordon” combines a deep-learning directional classifier — trained on nine features with quarterly retraining and out-of-sample validation — with a mean-reversion execution engine. Each trading day, the model forecasts each input parameter t+1 to generate a next-day direction signal; when that aligns with a mean-reversion entry trigger, Gordon autonomously enters or exits a fully-margined long or short position.
View project →Machine Learning · Finance
Credit Risk Classifier
Live loan default predictor built on a Random Forest classification model and deployed via Django on Heroku. Takes eight loan-level inputs — term length, amount, business location type, company age, borrower state/city, employee count, and issuing bank state — and returns a probability of default. Model was iteratively trained and pruned from a dataset provided by Dr. Brent Albrecht.
View project →