XAVIER CAPDEPON
github.com/chabir acux28@r.postjobfree.com 1-917-***-****
EXPERIENCE
Consultant – Kiski Alpha Partners, New York, NY 01/2016-Present
Created and developed specific prototype algorithm based on a proprietary SQL database to visualize, understand, quantify and verify the investment management style of various asset managers using several financial factor/ratio such as Momentum, Price-to-Earning, Price-to-FreeCashFlows.
Designed a script to extract the diverse components of an asset manager portfolio returns based on the Fama-French model.
Created scripts from scratch using Python, SQL and R, RODBC and visualizations using R Shiny package,.
Consultant - T. Analytics, (Big Data Consulting) New York, NY 09/2015 – 11/2015
Created and developed specific prototype algorithm, utilizing NLP (natural language processing), on a large unstructured social media data set (30+GB) from a community social network to discover trends in the fashion industry; script is written in Python, R, SQL, Spark and NLTK.
Created visualizations using networkX and networkD3 based on extracted data to facilitate decisions concerning the client business.
Consultant - Z. Family Office (Quantitative Data Analytics), New York, NY 07/2012 – Present
Built a script to automatically reconcile data for a large investment portfolio - reduced process time by 90% using VBA and Bloomberg Add-In
Wrote a script to format large amount unstructured data from PDF statements to be later reconciled with other databases.
Consultant - Open Data Enterprise, Washington, DC 08/2015 – 10/2015
Performed statistical analysis to analyze their proprietary data
Added relevant external data to generate more insight and comparison with similar organizations.
Implement various data visualizations that were included in a final written report and presentation to the staff.
Guggenheim Securities, Vice President, Structured Products, New York, NY 06/2008 – 07/2012
Built multiple tools in Excel/VBA/C#/Bloomberg/Intex APIs to analyze Credit Derivatives; Capabilities include: automated pricing of collateral bottom up based on market assumptions
Analyzed and traded Credit derivatives such as securitized and insurance products.
Improved infrastructure by implementing a SQL/VBA database containing key structural features, collateral quality covenants and risk/return sensitivity analysis for various securities
Created a parsing tool to clean (regex) HTML pages to provide industry news to sales forces
Developed an Excel/VBA BWIC CLO tranche pricing tool to automate process and efficiently analyze securities; Increased desk productivity by 30%.
Implemented an automated monthly risk and pricing valuation tool for 800+ securities; built in Excel/Access/Intex/C#/VBA to compile relevant data, compare key metrics and stratify deals based on key assumptions
Performed portfolio analysis using rating agency methodology to assist issuers with the development of an optimal structure; Generated cash flow and duration analysis; Modeled waterfalls depending on various sensitivity stresses
Structured VC backed ABS deals and CLOs using INTEX modeling software & Excel/VBA, including:
o$250MM of Middle Market CLO 2.0 closed in 2011
o$224MM of Broadly Syndicated/Middle Market CLO 2.0 closed in 2012
o$271MM of Venture Debt (life science and health care loans) Esoteric ABS closed in 2011
o$129MM of Venture Debt (health care loans) Esoteric ABS closed in 2012
Mizuho Securities, Associate Vice President, Structured Credit, New York, NY 12/2006 – 05/2008
Reverse engineered existing CDO transactions using INTEX script (C#) ~ $5BN +; repackaged and rated them using ratings methodologies (Moody's Correlated Binomial Distribution & S&P Evaluator)
Responded to various business and management questions to ensure deal closings and coordinated inquiries /problem resolutions with all necessary parties of the transaction (lawyers, auditors, trustee, investors, etc)
Credit Agricole Investment Bank, Associate, Structured Credit, New York, NY 01/2005 – 12/2006
Modeled eight cash CDO transactions ~ $10BN +
Created Excel/Intex script financial models to optimize cash flow waterfalls for rating arbitrage structures, hybrid CDO deals of cash and synthetic ABS
Mercur, Research and Innovation Dept. VEOLIA, Public Transport Subsidiary, Paris 04/2003 – 09/2003
Optimized bus lines timing; implemented simulations using Monte Carlo simulation - C++/Excel/VBA; budget: .6MM €
EDUCATION
NYC Data Science Academy, New York, NY 06/2015- 09/2015
Classes: Numerical & Statistical Computing; Machine Learning; Data Mining
E.M LYON, Graduate School of Business and Management, Lyon, France 09/2003 - 12/2004
MS in Corporate Finance
EIVP, Ecole des Ingenenieurs de la Ville de Paris, Engineering School, Paris, France 09/1999 - 06/2002
MSc in Civil Engineering
QUALIFICATIONS/LEADERSHIP
Certifications:
GARP FRM Program (Financial Risk Manager) Exam Part I (2014);
SOA/CAS (Society of Actuaries and Casualty Actuary Society) Exam 2/FM & Exam 1/P (2014 &2015)
General Computer Skills: Data Science Prog.; Machine Learning; Parallel processing; MS Office; VBA; C#/C++; Intex, Bloomberg API
Statistical Programming & Data Processing: R, Python, SQL, Spark, Hadoop, pandas, numpy, scikit learn,
Visualization: ggplot2, Shiny, seaborn, matplotlib, networkX, networkD3
Methods: General Linear Model, Logistic, SVM, Tree-Based Models, Bayesian Methods, Boosting, Dimensionality Reduction, Unsupervised & Supervised Learning
Languages: Fluent in French & English
DATA PROJECT WORK & KAGGLE COMPETITION
“Most popular R packages and R packages dependency visualizations” – R, Python, Spark, pandas, numpy, networkD3
Find the most popular downloaded R packages using the sparse matrix concept and paralleling processing (Spark) using the CRAN R packages description and package download logs websites
“Fastest subway in NYC (Data updates every 10seconds!)” – R, Shiny App, Python, ggplot2
Study the travel time from Yonkers to downtown Manhattan with the local 1 and express 4 trains using Shiny App visualization tool; Real time subway map visualization
Kaggle Competitions – Current rank: 812th / 500,000+ data-scientists -- world’s biggest predictive modeling online competition platform
*33rd /2926 - BNP Paribas Cardif Claims Management
Data: fully anonymous variables & binary target -- Solution: extensive feature engineering and ensembling methods
*40th /2619 – Prudential Life Insurance Assessment
Data: semi-anonymous var. & ordinal scaled target – Solution: feature engineering & customized algorithm for ordinal scale
ACTIVITIES/INTERESTS
Interests: Pastry & baking, hiking, home DIY, Kaggle online competitions (812th over 520,000+ participants)
Speaker:
Guest Lecturer at NEOMIA Business School: “Financial Securitizations before and after the 2008 crisis”
Guest Speaker for the NYC Open Data Meet-up: “Shiny App Demo: Fastest subway in NYC” & “Use python and Spark to dig into 2.5GB/47millions lines of log reports packages” (Aug. & Sep. 2015), & “Kaggle competition initiation” (Jan. 2016)