Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 132 views

    Arif Jahangir: Data Scientist - Machine Learning Engineer

    Uploaded by

    jani
    Arif Jahangir is a data scientist with expertise in machine learning, deep learning, computer vision, NLP, and predictive modeling. He has over 15 years of experience delivering data-driven solutions to business problems. Some of his accomplishments include creating consumer behavior models for Accor hotels using BERT, building anomaly detection algorithms for Ryerson University, and developing a stock prediction algorithm using graph networks. He is proficient in Python, R, SQL, and machine learning libraries.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd

    Arif Jahangir: Data Scientist - Machine Learning Engineer

    Uploaded by

    jani
    132 views4 pages
    Arif Jahangir is a data scientist with expertise in machine learning, deep learning, computer vision, NLP, and predictive modeling. He has over 15 years of experience delivering data-driven solutions to business problems. Some of his accomplishments include creating consumer behavior models for Accor hotels using BERT, building anomaly detection algorithms for Ryerson University, and developing a stock prediction algorithm using graph networks. He is proficient in Python, R, SQL, and machine learning libraries.

    Original Description:

    data

    Original Title

    Arif-DATA

    Copyright

    © © All Rights Reserved

    Available Formats

    DOC, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Arif Jahangir is a data scientist with expertise in machine learning, deep learning, computer vision, NLP, and predictive modeling. He has over 15 years of experience delivering data-driven solutions to business problems. Some of his accomplishments include creating consumer behavior models for Accor hotels using BERT, building anomaly detection algorithms for Ryerson University, and developing a stock prediction algorithm using graph networks. He is proficient in Python, R, SQL, and machine learning libraries.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd
    Download as doc, pdf, or txt
    132 views4 pages

    Arif Jahangir: Data Scientist - Machine Learning Engineer

    Uploaded by

    jani
    Arif Jahangir is a data scientist with expertise in machine learning, deep learning, computer vision, NLP, and predictive modeling. He has over 15 years of experience delivering data-driven solutions to business problems. Some of his accomplishments include creating consumer behavior models for Accor hotels using BERT, building anomaly detection algorithms for Ryerson University, and developing a stock prediction algorithm using graph networks. He is proficient in Python, R, SQL, and machine learning libraries.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd
    Download as doc, pdf, or txt
    You are on page 1of 4

    Arif Jahangir

    Mississauga, ON L5C2V7 | +1647 915 5975| arifjahangir@zoho.com |


    https://www.linkedin.com/in/arifjahangir

    https://github.com/arifj999

    Data Scientist – Machine Learning Engineer


    Communication – Predictive Modeling- Computer Vision- Text Analytics- Anomaly
    Detection - Stock Price Prediction-– Financial Forecasting - Project Management

    Providing data-driven, action-oriented solutions to challenging business problems


    Business-minded data scientist with a demonstrated ability to deliver valuable insights via data
    analytics and advanced data-driven methods. Accomplished communicator skilled in collaborative
    relationships, conveying big-picture insights, and driving opportunities through team support and
    success. Strong knowledge and comfort with visualization and communication relating to mathematics,
    machine learning, deep learning, and applied statistics. Skilled in trend and relationship identification,
    interpretation, and translating information into actionable strategies for growth. Highly adaptable
    professional with extensive experience in developing and deploying tools, software, and reports aimed
    at improving user data access and analysis. Worked in an Agile environment.

    Expertise

     Analytics  Graph-Nets  Data Structures


     Complex SQL  Computer Vision  Data wrangling of
    Queries  NLP structured,
     Computational  Reinforcement unstructured ,
    Statistics Learning messy and missing
     Machine Learning  Optimization of data
     Deep Learning Algorithms
     Anomaly  Strategic Thinking
    Detection  Data Integration
    Professional Experience

    RYERSON UNIVERSITY Data Scientist, 2015 to Present


    Managed all aspects of projects from conception to execution to deployment, delivering
    relevant scientific and business results
    Selected Achievements:

    A project for the Accor hotel group, a chain of 4000 hotels around the world. Created consumer behaviours
    models by exploiting the sequence and attention nature of Bert architecture. Customized Bert architectures
    were trained on more than 100 million records by using spark framework.
    Constructed anomaly detection algorithms by representation learning for reconstruction using various
    variations of Principal component analysis (PCA) , Autoencoders (AEs), Convolutional Auto-Encoders (CAEs),
    Contractive Autoencoders, De-noising Auto-Encoders (DAE) and Stacked DAEs (SDAEs). Constructed anomaly
    detection algorithms by predictive Modeling .using Bidirectional LSTM, GRU and Convolutional long short-
    term memory (ConvLSTM) , Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), and
    Adversarial Autoencoders (AAEs) with attention mechanisms.

    Time Series Classification by Augmenting Multi-Scale Convolutional Neural Network with Recurrence Plots,
    Gramian Angular Plots And Markov Transition Fields. In this research, we combine MCNN with the encoding of
    time series as images of Gramian Angular Field (GAF), the Markov Transition Field (MTF) recurrence plot into
    the signal deep convolutional neural network. The network will primarily be used for classifying internarial
    pressure (ICP).

    Created stock prediction algorithm by using latest graph-net architecture and latest Bert-Transformer
    architecture by exploiting the inter-dependencies of all stock on each other. On the basis of results created
    different portfolios for investments that optimize return on investments.

    Worked and implemented various architecture of convolutional neural network (CNN) like ResNet, AlexNet,
    VGGNet, Inception, DeepDream, Neural style transfer, Face recognition, Object Recognition (YOLO & SSD) and
    Pose Estimation.

    Please see my LinkedIn profile for details

    WYSDOM.AI Richmond Hill Data Scientist Nov 2017 – Nov 2018

    Created various algorithms by using latest Bert-Transformer architecture and its derivatives algorithms for
    similarity, text classification, inference and neural machine translation(NMT). Worked and implemented the
    various architecture of recurrent neural network (RNN) like LSTM, GRU, Encoder-Decoder models, Neural
    Machine translation, Sequence 2 Sequence, Build language models, text classification, Question Answering
    System, text summarization, entity extraction algorithms

    Please see my LinkedIn profile for details

    MISHAL (PVT) LIMITED Data Scientist July 2009 – Aug 2015

    Directed a team of professionals in the delivery of innovative projects for global impact. Planned and
    executed projects for societal health and product development using targeted analysis techniques,
    including modelling, forecasting, machine algorithms, and computational statistics. Developed software,
    STEEEPA trends, and processes for client companies
    Selected Achievements:
    Planned, Organized, Modeled, Standardized, Projected, Visualized health of a society. Various blood tests
    were taken from 1500 individuals this was treated as a sample (just like statistics in the polls), and
    clustering algorithms and propensity score analysis were applied to segment space generate by features
    of blood tests and health of the society were referenced by computational statistics and clustering
    algorithms. This project was done yearly

    Conceptualized, Analyzed, Forecasted, Tabulated, Marketed and discovered product categories. The
    periodic survey was conducted. Space was built of the dimensions of demographic & socio-economic
    indicators of individuals. Segmentation (classification) of this space was done on product category by
    using support vector machine and deep learning algorithms. This was done for every product category,
    and it was done iteratively. This work resulted in strategic target marketing, which gave 40-55% boost in
    revenue of client companies.
    Developed software to automatically detect fault lines in X-Rays. The software was based on
    representation learning and deep learning.
    Please see my LinkedIn profile for details

    National Revenue Data Processing Center (NRDPC)-Assistant Manager-2001-to 2005

    National Revenue Data Processing Center (NRDPC) is the headquarter, where tax data from all over the
    country is collected, collated, and compiled. Using creative insights of the working of the organization
    and the environment in which it was working, processing time and communication time were reduced to
    40%, and cost of operation was reduced to 30% compared to what was done previously.

    Job Responsibilities at NRDPC included performing complex SQL queries on more than 10 million records
    to collect, collate, analysis and Interpret the tax database. Yearly preparations and analysis of reports on
    tax data (under-reporting, Calculation of Tax Potential, Tax profiles, etc.). These reports were sent to the
    Finance Minister, Prime Minister and International Monetary Fund (IMF).

    Quantifiable results:
    Directed a team of professionals through innovative projects that decreased both the cost and time
    requirements.
    In anomaly detection attain an accuracy of 90% to 99.5% on various datasets, which reduced 35% the
    cost of maintenance and 25% time of maintenance.
    Using creative insights of the working of the organization and the environment in which it was working,
    processing time and communication time were reduced to 40%, and cost of operation was reduced to
    30% compared to what was done previously
    Targeting marketing resulted in a 40-55% increase in revenue for client companies.
    Implemented visualization tools like D3, Tableau or QlikView. Proactively identify opportunities to
    enhance/build reports and dashboards to elevate insights, which decreased the decision time by 45%.
    Created a deep learning card game https://youtu.be/W7BkXWjRA48
    Completed a project a year before the anticipated deadline.

    Education
    Ryerson University | Master’s in Computer Science (M.Sc.) specialized in machine learning and deep learning|
    GPA 4.2/4.33
    Quaid-e-Azam University | Master’s in Physics (M.Sc.)
    COMSATS | Postgraduate diploma in Information technology
    Punjab University | Bachelor’s degree in pure and applied Mathematics (B.Sc.)

    Coursework
    Mathematics with Applications in Finance, MIT | Machine Learning and Artificial Intelligence, Stanford
    University | Natural Language Processing, Stanford University| Deep Learning, Toronto University
    Certifications
     Structuring Machine Learning Projects by deeplearning.ai on Coursera. The certificate earned on
    August 26, 2017
     Improving Deep Neural Networks: Hyper-parameter tuning, Regularization, and Optimization by
    deeplearning.ai on Coursera. The certificate earned on August 25, 2017
     Neural Networks and Deep Learning by deeplearning.ai on Coursera. The certificate earned on August
    23, 2017

    TECHNICAL PROFICIENCIES

    Languages and Softwares:: R; Python, Keras, numpy, scipy, sci-kit-learn, panda, theano,
    TensorFlow; Spark MLlib, MXNet , Matlab, SAS, SPSS, Microsoft Office Suite, Tableau, Microsoft
    Power BI, Qlik View, Mathematica, SQL Plus, Transact-SQL, Microsoft Access, Oracle Developer,
    PL/SQL. Linux stack(bash, git, package management etc.)

    Machine Algorithms / Statistical Knowledge: Bert-Transformer, Reinforcement Learning,


    Graph-Nets Combined with Deep Learning, ResNet, AlexNet, VGGNet, Inception, DeepDream,
    Neural style transfer, Face recognition, Object Recognition (YOLO & SSD), Pose Estimation (RNN)
    like LSTM, GRU, Encoder-Decoder models, Neural Machine translation, Sequence 2 Sequence,
    Build language models, text classification, image captioning. Clustering, Bayesian Networks,
    Market Basket Analysis, Multivariable Modeling, Decision Tree, Hierarchical Analysis, Support
    Vector Machine, Longitudinal Designs and Repeated Measures, Neural Networks, Prediction
    Modeling, PATH Analysis, Anomaly Detection, Recursive Partitioning, Associations Rules,
    Multiple Imputation and Deep Learning, Recommender System, Propensity Score Analysis,
    Representation Learning, and Wavelet Analysis.

    You might also like