Machine Learning
Credit-risk models shipped to production, with Bayesian tuning, validation, and segment-level error analysis.
/ Quantitative ML · Interpretability · Agents
ZachariahFarahany
I build, deploy, and research production ML and agentic systems.
◇ Capabilities
Generative & Agentic AI
RAG systems, multi-tool agents, and MLLM evaluation, plus LLM interpretability research.
ML Operations
Automated deployment, testing, and monitoring of models on cloud infrastructure, with resource optimization and MLFlow-based management.
Big Data
Terabyte-scale ETL on Spark and medallion pipelines on AWS, taking raw telemetry through to production feature stores and warehouses.
→ Selected work
- 01Machine Learning
Predicting COVID-19 Length of Stay
Neural-network length-of-stay prediction on high-dimensional EHR data, and the basis of an IEEE Big Data conference paper.
ANN · SMOTE · Bayesian Optimization - 02Robotics
SkySearch UAVs
Open-source, multi-modal autonomous drone search with open-vocabulary target identification driven by multimodal LLMs. My MS thesis; 95% live-flight success.
Python · Computer Vision · MLLM · DJI Tello - 03Generative AI
Chartwell Insurance Assistant
A production RAG-powered email-automation assistant that answers Chartwell’s most frequent support questions.
GPT-4 · Pinecone · RAG
