Dhruv Madeka
Data AnalystMember of Technical Staff at Anthropic
Anthropic
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Member of Technical Staff at Anthropic
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
New York University
Data Science / Machine Learning
January 1, 2016 – Present
Questrom School of Business, Boston University
Master of Science (MS), Financial Mathematics
January 1, 2012 – January 1, 2014
University of Michigan
B.E, Industrial And Operations Engineering
January 1, 2008 – January 1, 2012
Anthropic
Member of Technical Staff
March 1, 2026 – Present
San Francisco, California, United States
Google DeepMind
Research Engineer
June 1, 2024 – February 1, 2026
New York City Metropolitan Area
Amazon
Principal Machine Learning Scientist
September 1, 2020 – June 1, 2024
Amazon
Senior Machine Learning Scientist
June 1, 2017 – September 1, 2020
Bloomberg LP
Quantitative Researcher
April 1, 2014 – June 1, 2017
New York City Metropolitan Area
Citizens Bank
Model Risk Validation Quantitative Research Intern
June 1, 2013 – August 1, 2013
Boston, Massachusetts
Optimal Asset Allocation in a Dynamic Factor Model Framework
November 1, 2013 – Present
Factor Models are the primary basis for tactical asset allocation amongst funds and institutional investors. They have largely been used in a static sense, with allocations based on myopically optimal weightings along risk factors. Here we suggest a method for optimal portfolios in a dynamic sense, with a factor structure for the spanned risks in the market. We use the tools of Malliavin Calculus to express the factor risks as conditional expectations of the State Price Density and the Market Price of Risk. The model creates an intertemporal hedging demand for the risk factors, which is also studied.
Estimating Covariance Matrices for Investments whose Histories Differ in Length
May 1, 2013 – Present
The Team developed a method to analyze investments with variable histories by imputing data from the joint conditional distribution of the returns from stocks in the S&P 100 Index. Compared and contrasted the estimates of the variance-covariance matrix generated by Principal Components Analysis, Asymptotic Principal Components Analysis and Naive Estimation. Developed an analytical solution for an unconstrained minimum variance portfolio as well as a rolling cross-sectional regression to determine minimum variance portfolio weights dynamically. A Backtest of the minimum variance portfolio showed that the naive estimates of the variance covariance matrices generated by the imputed data matrix performed the best out of sample. Work in Progress:Considering methods in Randomized Matrix Theory to determine number of eigen values for Principal Components Analysis of imputed data matrix. Study into non conjugate pairs for dynamic update of non-normal conditional distribution of missing data.
Judging Compensation Based On Performance Of Large Cap Bank CEOs
December 1, 2012 – Present
The relationship between CEO compensation and performance over the period 2000 - 2012 for the 19 large banks stress tested by the Fed will be investigated. A rating system will be developed to help a board of directors judge if the CEO is overcompensated. My task will be to find data, decide the variables that determine performance of CEOs and assist test for the existence of correlation between compensation and performance and develop a vector autoregression model which can predict compensation based on chosen performance indicators.
Cultural Fit Analysis
The candidate's background in quantitative finance and machine learning, coupled with experience at leading research and tech institutions, suggests a strong fit for roles requiring rigorous analytical thinking and innovation. The personal projects demonstrate initiative and a deep interest in complex data-driven problems. The transition from financial mathematics to data science and machine learning indicates adaptability and a continuous learning mindset, which are positive indicators for cultural fit in a dynamic environment. The target role of 'Data Analyst' might be a slight mismatch given the candidate's Principal ML Scientist and Research Engineer experience, suggesting they might be overqualified or seeking a more specialized/advanced data science role.
Soft Skills & Operational Fit
The candidate's project descriptions indicate a strong analytical mindset and the ability to tackle complex, multi-faceted problems. The progression through senior and principal roles at major tech companies suggests leadership potential, strong problem-solving skills, and the ability to operate effectively in demanding environments. However, without specific behavioral or psychometric test results, a detailed assessment of soft skills like teamwork, communication style, or stress handling is not possible.