Disha NEET Physics Guide for classes 11 and 12.pdf
Resume
1. PARMANAND SAHU
Dallas,TX | parmanand.sahu@utdallas.edu | 972-730-3967 | linkedin.com/in/parmanandsahu/ | https://parmanandsahu.com/
EDUCATION
THE UNIVERSITY OF TEXAS AT DALLAS, Richardson, TX Aug 2019 - May 2021
Master of Science in Computer Science 3.66/4
NATIONAL INSTITUTE OF TECHNOLOGY, Raipur, IN Jul 2009 - Jul 2013
Bachelor of Technology (Hons.) in Metallurgical Engineering 8.35/10
TECHNICAL SKILLS
Languages Python, C, Java, Scala, Matlab, Node.js
Databases MongoDB, MySQL, DynamoDB, Neptune, ElasticSearch, Neo4j
Libraries Numpy, Pandas, Matplotlib, Plotly, NLTK, Gensim, Sklearn, Spacy, Scipy, Tensorflow, PyTorch
ML Algorithms Logistic Regression, Dimensional Reduction, SVM, Clustering, Tree Based Algorithms, Ensemble Techniques
DL Algorithms RNN, LSTM, CNN, Attention Mechanism, Word Embeddings
NLP task Sequence-to-Sequence, Sequence tagging and Classification, Named Entity Recognition, Question Answering
Big Data Tools Hadoop, Spark, PySpark, MLlib, Hive, Impala, GraphX
Technologies Linux, Git, Django, Rest-API, Flask, Docker, Kubernetes, ML-flow, AWS, ECS, EKR, ECR, Nextflow
WORK EXPERIENCE
Lantern Pharma : Clinical stage pharmaceutical company Jan 2021 - Apr 2021
Data Scientist and Platform Development Intern Dallas,TX
Data Pipeline for Cancer Drug exploration platform
– Develop module to ingest genomics data (TCGA) of 50k+ samples and 3 billion+ data points into cloud infrastructure
– Designed and implemented microservice in AWS for ingesting genomics data
Capital One: Bank Holding Company Jun 2020 - Aug 2020
Data Science Intern McLean,VA
Neural Network Model to predict Mortgage Based Security Prepayment Rate
– Performed EDA and preprocessesed a specific category of Mortgage Based Securities(MBS) in investment portfolio.
– Implemented parallel processing for tuning hyperparameters of model to predict prepayment rate with extensive logging.
– Built and analyzed automated performance report (PDP/ICE, SHAP, and S-Curves) for comparing models.
VHSS Lab: Research Lab at UTD Sep 2019 - May 2020
Machine Learning Specialist Richardson,TX
NSF funded Conversational Emotive Virtual Reality patient project
– Researched and trained transformer-based model for virtual patient interacting with medical students using Pytorch.
Huddl.ai : Video Communications Provider Apr 2018 - Jun 2019
Artificial Intelligence Engineer Hyderabad, India
Named Entity Recognition for Voice Assistant
– Supervised data preparation team and retrained custom spacy model for Named-Entity-Recognition..
– Built module using Levenshtein Distance and Phonetic similarity to fix incorrect transcription for recognized entities.
– Developed micro-service using Node.js and DynamoDB to use as gazetteer in NER.
– Designed module using regex for extracting entities like Time and Date from voice commands.
– Packaged micro-services into docker for deployment in Kubernetes.
Reverse image search for information retrieval
– Developed parser for OCR response and utilized K-means to classify content to reduce false positive.
– Built keyword extraction using RAKE and graph-based algorithm for ranking meetings on search results.
Action Item Detection in Meeting Transcript
– Trained LSTM-RNN based model to classify the action items in the meeting transcript 95% accuracy.
– Deployed ML-flow for internal use and track experiments with different hyperparameters.
CoArtha Technosolution: Talent Acquisition platform Sep 2017 - Apr 2018
Associate Data Science Engineer Hyderabad, India
Semantic Understanding of Job Description for ranking resumes
2. – Trained model using Naive Bayes to classify sentences in job descriptions with 90%+ accuracy for matching resumes.
– Assisted in developing scoring logic to match job descriptions with resumes.
Candidate screening from audio interviews
– Employed Random Forest for classifying candidates using interview response audio with 90+% accuracy.
CoArtha Technosolution: Talent Acquisition platform Sep 2016 - Aug 2017
Associate Software Engineer Hyderabad, India
Knowledge Graph from job descriptions for ranking resumes
– Built a part of pipeline to scrape US job boards and pre-processed data using Selenium and Beautiful Soup.
– Built a part of Knowledge graph using Neo4j with skills, job titles & education entities from 100k+ job descriptions.
Semantic parsing of resumes
– Trained ensemble model to identify sections(contact, education, experience and skill)
– Implemented solution using regex for extracting entities and parsed table to correlate extracted entities.
DigiFledged: Digital Marketing Startup Jul 2015 - Aug 2016
Founder Bhilai, India
– Managed daily operation and acquired technical & functional requirements of projects from new clients.
– Led a team to deliver 5+ web development, 17+ freelancing projects and establish a blog with 130K+ page-views.
JSW Steel Ltd: Steel Manufacturing Company Feb 2014 - Apr 2015
Junior Manager Bellary, India
– Analyzed production reports discovering insights through exploratory data analysis using MS Excel and R.
PROJECTS
Question Answering on SQUAD 1.0 (LSTM | RNN | Self Attention | Pytorch)
– Preprocessed and extracted custom features along with pre-trained word embedding(Glove).
– Trained QA model(simplified Stanford Attentive Reader) with 70% F1 Score.
Language Model for Auto complete sentence
– Preprocess data for n-gram language model with smoothing for sentence auto complete
Named Entity Recognition on CONLL 2003 (RNN | GRU | Pytorch)
– Preprocess and prepare vocabulary for embedding layer
– Trained and compared Vanilla RNN(83%) and GRU RNN(86%)
Document search using approximate k-nearest neighbor
– Implemented local sensitive hashing(LSH) for multiple universes (different set of random planes)
– Developed document search using approximate k-nearest neighbor and LSH
Credit Card Transaction Fraud Detection(Random Forest | Logistic Regression | Feature Engineering | Sci-kit)
– Imputed missing data,created custom features,normalize and encode features
– Performed exploratory data analysis of features
– Handled imbalanced data using SMOTE and custom loss function.
– Train and evaluate linear/tree based classifier methods
– Analyzed feature importance w.r.t to dependent variable
Image Identification on CIFAR-10 dataset (Pytorch | Convolutions Neural Network )
– Augmented image data using transformation technique(random crop,vertical flip)
– Implemented and trained RESNET family of architecture for image detection with 86% accuracy
Ensemble Method and Decision Tree from scratch (Decision Tree | Bagging | Adaboost | Sci-Kit)
– Implemented fixed depth decision(ID3) tree from scratch for monk’s classification dataset
– Implemented Bagging and AdaBoost and compared with Sci-kit implementation for Mushroom bruises
CERTIFICATIONS AND ACTIVITIES
– Natural Language Processing, Machine Learning and Deep Learning by Coursera
– AWS Services by The University of Texas at Dallas and Linkedin Learning
– Linked Data Engineering by Hasso-Plattner Institute: Building Knowledge Graph,2016.
– M101: MongoDB for Developers by MongoDB University