Generable is on a mission to bring more clarity to early drug development programs starting with oncology. We use the latest advances in Bayesian and causal Inference, machine learning, and probabilistic computing to answer questions about dosing, toxicity, patient and biomarker selection, and efficacy relative to competitive products and standard of care. We have a curated dataset of patient-level data and published research to inform our models. Generable originated from the open-source Stan (mc-stan.org) project at Columbia University.

Teaching Bayesian data analysis and a methods boot camp, including probability, linear algebra, and statistical computing.

Providing statistical consulting services to the sales team and to Spotfire clients. Working on interactive data and statistical model visualizations using Spotfire and R/S+. Occasionally presenting at industry conferences.

Building market and pricing analytics tools for the publishing industry. Performing statistical analysis and visualizations using R, ggplot, and SQL.

Performed statistical analysis of order flows to better understand profitability across order types and uncover opportunities for postion hedging and unwinding.Worked with the listed options desk to define requirements for high volume listed options trading system supporting electronic and high touch order flow, options spreads, and algos. Managed the roll out of the trading software to the floor brokers.

Risktail was part of the inaugural class of NYC SeedStart 2010 start-up accelerator.Together with two co-founders we built a tool for backtesting, simulation, and optimization of options portfolios. Later we developed algorithms for scenario analysis and integration with the R system and developed several methods for visualizing and quickly interpreting the results of backtesting.

• Built time series models for FX markets to understand the dynamics of daily price fluctuation, set risk limits, and produced forecasts. Simulated from the model to evaluate goodness of fit.• Evaluated classifier performance for predicting income categories from census data using Bayesian Logistic Regression, Support Vector Machine, and Neural Networks. Used opim() R routines (loosely equivalent to fminunc() in Matlab/Octave) to evaluate tradeoffs between various numerical optimization algorithms.Trading and risk management of liquid listed options, mostly in Indexes, ETFs, and Futures. Used predominately hedged theta positive positions and managed risk using beta weighted greeks with correlated indexes.

I managed software projects and quality for General Services Administration (GSA).

I managed systems integration projects and large IT outsourcing projects.