GISjobs.com – It All Starts Here

  

Debdas Ghosh

Pittsburgh, PA, USA

DEBDAS GHOSH
Pittsburgh, PA (Open to relocate) | (412) 320 0303 | deg107@pitt.edu
linkedin.com/in/ghoshdebdas | github.com/debdasghosh | kaggle.com/debdascmu | debdasghosh.com
EDUCATION AND CERTIFICATIONS
MS Information Science University of Pittsburgh, GPA:3.9 May 2021
MBA(Systems) West Bengal University of Technology 2018
BSc Computer Application University of Calcutta 2009
Intro to Deep Learning Carnegie Mellon University 2020
Data Science and Machine Learning Ivy Professional School 2017-2018
TECHNICAL SKILLS
Expertise: Machine Learning, Neural Network (MLP, CNN, LSTM, Transformers), Computer Vision, Natural Language Processing, GIS, Database
Languages: Python (PyTorch, TensorFlow with Keras, Tensorboard, SciPy, scikit-learn, NumPy, Pandas, Matplotlib, OpenCV), R, C++, SQL, shell scripting, PHP, Javascript, HTML, CSS, JQuery.
Technologies: AWS, Git, Linux, MS-Office, Tableau, ArcGIS, Hadoop, Spark, Docker, Google Cloud.
PROFESSIONAL EXPERIENCE
Teaching Assistant U. Pitt. School of Computing Sept 2020-Present
• Intro to Computer Programming, teaching a class of 96 students the fundamentals of Java programming
Research Assistant U. Pitt. Bioinformatics Lab May 2020 - Sept 2020
• Developed Explainable AI for Deep Learning models (DenseNet, GANs). Converted models to Pytorch from Tensorflow 1.0.
• Worked on Medical image processing pipeline, Image Segmentation (UNet), DICOM to NIFTI, NIFTI to Supervoxels conversion. Conducted PyTest unit testing of Lung Pipeline.
• Analyzed the SCCOR dataset (CT images of Lung) using our own visualization tool PyViz Med and SimpleITK, Nibabel, PyDicom among others.
Asst Manager, Manager – IT Spencer’s Retail Ltd, Kolkata Apr 2013 - Aug 2019
• Conducted data mining, data modeling using MySQL; statistical analysis, business intelligence gathering using Python; presentation and benchmarking using Tableau.
• Provided data analytic supports decisions for customer service improvement, stock fill rate approximation & process reengineering resulting sales up by 12%, customer satisfaction by 18% and footfalls by 21% on YoY basis in 2018-2019.
• Extracted quantitative data to develop an understanding of customer behavior, demographics and buying patterns from MySql, SQL Server, SAP BW databases.
• Realized SCM Analytics, sale predictions, Market research analytics and Exploratory Data Analytics (EDA).
• Dealt with time complexity, optimization, feature selection, data extraction, data cleaning, Dashboard Reporting, Presentation, Hypothesis Testing.
ACADEMIC PROJECTS
Generating Accurate Human Face Sketches from Text Descriptions using DCGAN and CycleGAN in Pytorch and CelebA dataset. (https://github.com/jeanmichaeldiei/11785-Text-to-Sketch)
Cell Nuclei Detection and Segmentation - Detect Cell Nuclei of different cell types and segment them using Mask RCNN model from biomedical images. (https://github.com/debdasghosh/Cell-Nuclei-Detection-and-Segmentation)
Ensemble of Self-supervised Classification - image classification by ensembling different algos of self-supervised methods of feature selection using un-labeled data then pass that model in downstream supervised models.
Frame Level Classification of Speech - identify the phoneme state label from raw mel spectrogram frames.
Face Verification Using Convolutional Neural Networks - given two images as input, output a score that quantifies the similarity between the faces in these images.
Transcript Generation - predict the phonemes contained in utterances using nonaligned train set.
Transcript Generation with Attention - transcribe a given speech utterance to its corresponding transcript
ACTIVITIES
Vice President – School of Computing and Information Graduate Student Organization, UPitt, May 2020

Debdas Ghosh

You must be a registered/logged in employer to contact job seekers.