Generative AI
-
Flashpoint: Efficient AI Document ComparisonA novel efficient and thorough entity-based comparison algorithm for large documents.
-
RAGs to Riches: Zero-Shot ClassificationExploring RAG systems for zero-shot multi-label classification of large documents.
-
James Webb Images GANGenerative adversarial networks trained to produce images in the style of James Webb Telescope photography.
-
Prompt Optimization with DSPyExploring Stanford's DSPy library for automated prompt optimization and LLM pipelines.
Machine Learning
-
Capturing a Heartbeat from an ImageHow I extracted the sound of my son's heartbeat from the image of his first ultrasound.
-
ACNH Turnip Stalk MarketTime-series clustering to predict turnip market trends in Animal Crossing: New Horizons.
-
U.S. Senate PolarizationClustering analysis to quantify divergence in U.S. senators' voting records over time.
-
Hazardous Freight ClassificationSequence modeling and embeddings to predict railcar contents from route data, NLP for supply chain.
-
Automatic Change Point DetectionNew approach for automating change point detection with cumulative sum differencing and Kneedle.
-
LSTM Anomaly DetectionLSTM autoencoders for detecting bad bearings in motor assemblies from vibration sensor data.
-
TF-IDF Document SearchBuilding a fuzzy search algorithm over project notebooks using TF-IDF.
Data Engineering & Simulation
-
Retail Sales SimulationSimulating individual customer behaviors to build a high-granularity business model from the ground up.
-
Senate Voting History ETLA MySQL database of historical senate votes, built with web scrapers and automated bots.
-
Hamlet NLP Spark PipelineUsing Apache Spark to create a practical NLP dataset from Shakespeare's Hamlet.
Dashboards & Analytics
-
Retail Simulation DashboardCustomer-level, seasonal, and regional sales trends from the retail simulation, built in Power BI.
-
Project Management DashboardSimulated project visibility, timelines, goals, and budget tracking.
-
Sales LeversClean display of key sales levers on a simulated dataset.
-
Retail Flight RiskA new statistical approach to measuring the impact of churn in retail settings.
Open Source
-
Bonsai LearningBonsai is an open-source, standalone, locally-hosted application for self-directed learning on any subject.
-
DSPy Dataiku AdapterPython-based adapter that enables Dataiku users to run DSPy's analysis, teleprompting, optimization, and agent workflows using Dataiku-native LLM Mesh models.