Career Profile
I am a Master Student majoring in Electrical Engineering and Computer Science with expertise in Machine Learning. My dissertation is on ‘Adversarial Attacks as a Threat to Machine Learning Models’. I’m looking for full time jobs in Machine Learning or Data Science to be started upon my graduation in May, 2019.
Projects and Experiences
- Adversarial examples capable of fooling image classifiers generated ex. adversarial examples to fool MNIST classifier, examples to fool finetuned VGG16 in CIFAR-10).
- Provided an observational and theoretical justification that a change orthogonal to the feature space can lead to adversarial examples.
- Wrote codes and detailed explanations on state-of-the art adversarial example generation mechanisms including: Fast Gradient Sign Method (FGSM), Jacobian-Based Saliency Map Method, and Carlini-Wagner Method.
- Evaluated the performance of the defense mechanisms of: Adversarial Training and Defense Distillation.
- Participated in 10 different competitions
- Vision: Developed image classifiers using transfer learning. In particular, finetuning ResNet-50 for competitions including ‘Dog vs. Cat’ (achieved accuracy: 99.7%, training time: less than 10 minutes)
- Structure Data (Rossmann Store Competition): TriedRegression analysis, Random Forest, Gradient Boosting Trees. My proposed fully connected NN with capability of learning embeddings for categorical variables reached a loss value at rank 10.
- Time Series: Performing time-series analysis through NN by supplying time features into the model as an approach easier than the traditional statistical time series models.
- Helping the quality department with creating quality inspection plans, Non Conformance Reports (NCR’s), \Corrective Action Report (CAR’s), Supplier Corrective Action Report (SCAR’s) and Supplier Submittal Request (SSR’s), Creating First Article Inspection (FAI) plan per QE’s direction
skills
Programming Languages - Python, SQL, Cython, C, C++, MATLAB, Mathematica
Numerical & ML Packages - NumPy, SciPy, Pandas, StatsModels, Matplotlib, Seaborn, Tensorflow, PyTorch, scikit-learn, Jupyter, XGBoost
Computer Vision - Convolutional Neural Network, Pretrained Net (VGG16, VGG19, ResNet, Inception-V3), Transfer Learning
Natural Language Processing - Recurrent Neural Network/LSTM/GRU, Word Embedding, Word2Vec, GloVe, N-gram, TF-IDF
Supervised Learning - Linear Models, Generalized Linear Models, Random Forest, Gradient Boosting Tree, LASSO, Naive Bayes
Unsupervised Learning - K-means, DBSCAN, HDBSCAN, KNN
Data Dimension Reduction - SVD, PCA, Random Projection
Cloud Services - AWS EC2, AWS S3, Paperspace