Data Scientist Adjunct Lecturer

Dalila Benachenhou

Washington, DC ***** 202-***-**** ********@*****.***

linkedin.com/in/dalila-benachenhou/ femvestor.blogspot.com github.com/dalilar

Data Scientist

Experienced in applying critical thinking, problem solving, and analytical skills to extract actionable insights from industry data. Industries have included oil and gas, retail, manufacturing, insurance, banking, finance, healthcare, biotech, non-profit, and federal agencies. Hands-on experience in using natural language processing (NLP), graph networks, artificial intelligence (AI), Machine Learning, and Deep Neural Networks using various languages and frameworks. Recognized for ability to develop creative solutions to hard problems, increase client-account renewals and up-sells, and produce results within expected time frame. Possesses ability to thrive in a collaborative environment involving cross-functional stakeholders and subject matter experts.

Skills:

Languages:

Python, SQL, R, MATLAB, Mathematica, Java

Machine Learning:

Random Forest, KNN, Linear Regression, Logistic regression, SVM, Naïve Bayes, Boltzmann Machine, Hidden Markov Model, xGboost

Platform:

DataRobot, Databrick, H2O, Palantir Foundry, Gitlab

Deep Learning:

Feedforward NN, Convolutional NN, Recurrent NN, Autoencoder

Packages:

Keras, Tensorflow, Trax, MxNet, caret, ggplot, pandas, numpy, NLTK, Spacy, AllenNLP, DataRobot

Big Data Tools:

Snowflake, Spark SQL, Paxata

Databases:

MySQL, Snowflake, postgresSQL, MongoDB, Redis

Time Series:

Stochastic Processes, AR, Garch, Extreme Distributions

Graph Networks:

Communities Detection, Small-World Networks, Metrics

Visualization Tools:

Tableau, Shiny App

Repository:

Git / GitHub

Notebooks:

Zepl, Zepplin, Google Colab

Others:

API, REST

Work Experience

DATAROBOT, Washington, DC July 2019 – September 2022

Execution Data Scientist

Drove end-to-end AI solutions with teams in various industries.

Moved 30+ use cases from framing to production.

Developed and shared modeling concepts and solutions for customer facing data scientists (CFDS) to drive adoption and implementation beyond personal use.

Helped CFDS with proof of concept to acquire new clients.

Mentored junior CFDSs.

Worked directly with clients to solve the more pressing and difficult cases.

Played key role in up-selling to federal and commercial companies.

Improved DataRobot’s ROI on at least two projects by finishing the projects before the deadline.

Provided advice to CFDSs on how to solve difficult cases.

Wrote blogs and answered questions from DataRobot community.

Worked with large and small data.

In federal agencies, helped increase account renewals and won new accounts.

Led three project teams. Conducted regular meetings to keep both customers and team members aligned and updated throughout the project life cycle to ensure projects’ success.

WORLD BANK, Washington, DC April 2018 – June 2019

Consultant

Developed customs fraud detection (Machine Learning and Deep Learning) models for Algeria and put it in production. Used R, Mathematica, and SQL.

Trained group in R programming and in Machine and Deep Learning Modeling.

Presented the work in French to Moroccan Customs officials, which won the World Bank a four-month contract.

Trained and mentored two university students to contribute to the Algerian Customs project.

As sole data scientist, built a prototype for Moroccan’s Customs and managed project from end-to-end.

GEORGE WASHINGTON UNIVERSITY, Washington, DC January 2013 – May 2019

Statistics Adjunct Lecturer

Taught third- and fourth-year students SAS, R, statistics, report writing, and best practices in getting value from data.

Revamped the syllabus and the class experience.

Received 92% approval rating from students.

Wrote at least one letter of recommendation for students per term.

WORLD BANK, Washington, DC February – May 2016

Consultant

Developed a framework to detect fraud and anomalies in Moroccan Customs.

Advised on use of machine learning and predictive modeling in anomaly detection.

The framework became used as a road map for further implementations.

FEMVESTOR, Washington, DC January 2005 – June 2019

President

Worked with hedge funds to build algorithmic trading models.

Developed due-diligence technology and was hired to perform research for an international not-for-profit organization. Based on results, organization decided to hire the executives researched.

Built US Financial Market model, within 10 days, for a wealth management start-up client whose team struggled for six months unsuccessfully to do so. Model enabled retirement simulations for client advising.

Built and maintained a self-updating website for 12 months, using Shiny app, to show new and historical construction in DC.

Education

SQL & AI Nanodegrees, Udacity, Mountain View, CA (online education)

Natural Language Processing with Attention Models, DeepLearning.AI (online education)

Natural Language Process with Classification and Vector Spaces, DeepLearning.AI (online education)

Master of Science, Statistics, American University, Washington, DC

Thesis: An Analysis of Methods for Fuzzy Rules Extraction

Master of Science, Computer Science, American University, Washington, DC

Thesis: PCR Primer Selection Using Expert Systems and Neural Networks Techniques

Bachelor of Science, cum laude, Computer Science, American University, Washington, DC

Projects

Automated Washington DC DPW 311 1 to 7 Days Forecast (May 2022 – July 2022)

Built end-to-end automated app on DataRobot platform to weekly forecast and retrain DC DPW 311 requests by Zip Codes, and publish the results on Tableau dashboard.

Python and R on Zepl notebook to schedule and access current datasets through API and update local ones.

DataRobot AI Catalog to store and clean datasets.

DataRobot MLOps to forecast and retrain model weekly.

Snowflakes to store final datasets for Tableau dashboard.

Resume Parser and ML model to match candidates to a job (January 2021 – June 2021)

NLP project.

Built a parser that extracted pertinent factoids from resumes.

Built NER for universities degrees and date parser.

Used Spacy, NLTK, AllenNLP, OCR, and pdf parsers.

Predict Engines Failure (June 2020 – August 2020)

Predict engines failure five days prior their failure.

Designed novel approach to detecting failure.

DataRobot AutoML to build models.

R to prepare datasets.

Tableau for results visualization.

Detect Customs Frauds (April 2018 – July 2019)

Built RNN, and Machine Learning models (supervised and unsupervised) to predict fraudulent imports.

While the occurrence of fraud in the dataset was less than 1%, for the fraud class, the recall was over 95%, and a precision was 50%. For non-fraud, the recall and precision were over 99%.

Putting the model in production.

SQL to prepare data for modeling.

Deep Learning to Predict Device Failures (December 2017)

Built Deep Feed Forward Neural Network (FFN), Random Forest, and KNN models for office room occupancy with what appeared severe class imbalance.

The FNN model reached an accuracy of 98.44%. Used MxNet to build it.

Mathematica.

Finding Connections Between Ingredients (August 2015 – July 2016)

NLP, graph network project.

Scraped over million recipes, uploaded them to AWS MySQL, and performed complex transformations, and high analytics to find affinities between ingredients in Mac N' Cheese recipes.

Built predictive model to understand recipes reviews. The model had an accuracy of 91.48%.

Python, Mathematica, SQL.

Executives and Companies Profiling (January 2011)

Built methodology and tools to speed up companies and executives profiling from public unstructured data. In the process, built in house graph database. The tool helped even profile executives from African countries. Check here for some examples. The project required dealing with many megabytes of documents and extracting relevant information that were then represented into a social graph network. Code written in Java.

Java, Lucene, graph network.

Additional Activities

Speaker on AI ethic and NLP.

Published papers and books on trading with Machine Learning, trading, and financial fraud detection.

Contact this candidate