Thierry Bema is a 44-year-old French national with experience in data science and enterprise business intelligence projects using technologies like Spark, Hadoop, and Scala. He has created multinode clusters on Hadoop and run various financial market and machine learning applications including stock price prediction, portfolio risk calculation, and sentiment analysis. His background also includes 17 years of experience designing RFICs for wireless technologies.
1. Thierry Bema
Age 44
Nationality :French
Address : 3 rue Crespieres
78580 Les Alluets le Roi (France)
Mobile : 0785866639
E-mail:thierry.bema@gmail.com
Bringing my full experiences in Enterprise oriented on BUSINESS
INTELLIGENT CORE UNIT gained after 17months of complete Achievements
of entire Scalable Projects in Data Science from Abstract Concepts to the
Real World solutions ( BIG DATA HADOOP YARN MESOS SPARK with the
Powerful Reactive programming Actor Akka Multithreading and Scala
Functional Programming for Machine Learning as being my key point
interests in our fast moving world of Big Data today including Graph
Processing, Streaming, SQL sustained by Python) which are based on my
self starter motivation attitude strongly inspired by AMERICAN SCIENTIFIC
CULTURE after a long career on Telecom Industry as being RFIC &
Microwaves Engineer
ARCHITECTURE IMPLEMENTATION
Creating and Building a Multinode Cluster (Ubuntu VMs) on Hadoop-2.7.2
with YARN and Spark1.6.0 Scalable through scripts based on Docker with
Serf Agent as the Orchestration communication between nodes
APPLICATIONS RUNNING ON THE CLUSTER (JFreeChart , Zepplin,
Breeze, Apache Common Math Libraries,Twitter scalding ,Akka,
Anaconda : IPython Notebook :NumPy,Scipy,Pandas,scikit-learn,h5py)
FINANCIAL MARKET PROJECTS :
Scalable Framework (Spark Akka Scala)
-Stock Price Prediction using full order from NYSE with Spark and Decision
Tree Algorithm for Classification (Spark Mlib :Multiclass Classification)
-Akka Cluster for Portfolio Market Risk Calculation which is depending on
factor forecasts (single node and multiple nodes) by using MonteCarlo
-Net Asset Stock Market Application Concurrent (with GUI) :
Taking a list of Stock ticker symbols along with the units of stock users hold
and tells them the stock value of their investments of the current date
UnSupervised Learning
-Extracting clusters from stock price action during a period by selecting the
appropriate features prior to clustering and time window to operate on by
using K-Means Algorithm
-Expectation Maximization (EM) Algorithm to estimate the maximum likelihood
observation based on posterior probability (stock prices from a couple industries movement)
OBJECTIVES
Starting
date :07/10 /2014
SYSTEM ENGINEER DATA SCIENCE
KEY ACHIEVEMENTS
2. Supervised Learning :
-Predicting the direction of change in the stock price movement based on
price index, change in Federal Fund rate and Gross Domestic Product by
using Naive Bayes Algorithm
-Computing of Single Variate Linear Regression of the price of Ticker
symbol over a period
- Predicting of price change between two consecutive trading session with
Multilinear Regression
-Prediction of positive price Variation for specific Ticker symbol given its
Volatility and Trading Volume by using Logistic Regression
Kernel Models and Support Vector Machines ( LIBSVM)
-Applying the Binary Support Vector Classifier to estimate the risk for a
company to eliminate its dividend based on realtime change in stock price
over the last few months (long term debt , Equity Ratio , Dividend coverage
ratio, Annual Dividend yield , Operating profit marging
Artificial Neural Networks (Multilayer Perceptron)
Understanding the correlation factors between Exchange rate of some
currencies ,the spot price of gold and S & P500 index (Extracting one or more
regressive models)
Genetic Algorithms :(Evolution Computing)
Chromosome genetic encoding of Trading Strategy , set of two signals to
predict the sudden relative decrease of price Security :relative Volume with
a condition and Relative Volatility with a condition
Reinforcement Learning (Q-Learning)
Option Trading , computing the best strategy to trade certain types of
options given some market conditions and trading data
VARIOUS PROJECTS IN DIFFERENT TOPICS
GraphX :( Healthcare)
-Query to track the source of Epidemic (Progression of Infection disease )
based on number of days exposure to other patient
Energy :
-Predicting Global energy demand using the available data the energy usage
on the last past years ,in order to enable companies effectively handle
energy demand (Using Brownian Motion continuos time Stochastic Process
by which drift in data is Standard Mean and Volatility in data is Standard
Deviation
AirLine : (on IPython notebook PySpark parallel distributed computation
accros the cluster on YARN )
-Computing the mean delay for take off and landing
Recommandation
Movielens data set
Classification Decision Tree
Titanic data set
Ending
date :17/12/2015
3. Project « LTE FONTEND RFIC 4G HARDWARE DESIGN » ( FREELANCE )
Project « Dual Band WLAN /WIMAX RFIC TRASNCEIVER) » (INFINEON:
INTEL MOBILE COMMUNICATION: Munich) VERIFICATION RF ENGINEER (
RFIC )
Project « UMTS BLUETOOTH » (FREELANCE)
Project « GSM DCS PCS » SAGEM ( SAFRAN )
Project « GSM DCS PCS » WAVECOMSA
Master in Microwave’s Circuits and Components ( University of
Montpellier 1998)
FINANCE ,GEOLOGY ,HISTORY , SPORTS (Running )
Starting date
07/2009/
Ending date
01/2014
Sarting date
11/2008
Ending date
08/2006
Starting date
6/2006
Ending date
9/2002
Starting date
11/2001
Ending date
7/2002
Starting date
8/2001
Ending date
1/200
RELEVANT EXPERIENCES MICROWAVES RFICS
DIPLOMA
INTERESTS