SANDEEP POLAVARAPU VENKATA NAGA
Machine Learning Engineer
College Park,MD
Summary
I help turn your data into decisions. I am seasoned ML engineer with a knack for using cutting-edge tech to tackle tough problems. I take messy data and turn it into a beautiful, scalable, and efficient machine-learning pipeline on embedded as well as cloud systems.
Overview
4
4
years of professional experience
4
4
years of post-secondary education
2
2
Certifications
Work History
Machine Learning Research Assistant
University of Maryland, College Park
02.2022 - 12.2022
- Implemented and trained a stacked Bi-directional LSTM for multi-classification of 12 crop types in a multi- spectral satellite image of Thailand as a Time-series forecasting problem
- Achieved producer accuracy of 96% and user accuracy of 92% (Github)
- Developed a data pipeline using DeepLabV3+ for semantic segmentation to assess the impact of sugar residue burning in the Thailand region, achieving an IOU (Intersection over Union) score of 77.7% (Github)
- Implemented parallel and distributed machine learning models on large data (10TB)
- Presented completed research projects at the NASA Land-Cover and Land-Use Change (LCLUC) conference (Link) under the guidance of Professor Varaprasad Bandaru
- Performed Numerical Optimization to fit a Double logistic function for Plant Phonology Estimation of wheat crop.
Machine Learning Engineer
TensorFlow | NLP | PySpark | AWS | ETL, Akamai Technologies
07.2018 - 08.2021
- Developed an ML-based solution for threat anticipation identifying Distributed denial of service (DDoS) attacks and BOT detection for critical events while ensuring fast and robust predictions on imbalanced data distributions
- Assessed models like Naive Bayes, Decision Trees, SVM, XGBoost, CNN-LSTM, and Autoencoders with A/B testing
- Periodically to stay current with changing threat patterns Online training was performed
- Business Impact: The predictions cut critical event reaction time by 30% with an accepted recall score of 82%
- Worked on a document search model development as a part of Search Akamai, a smart search engine designed to search for relevant documents within the company’s database optimally using LDA (Latent Dirichlet Allocation) and vector space models
- Additionally, Word2Vec skip-gram embeddings, K-Means and Hierarchical Clustering were experimented for Topic Modeling
Education
Master of Science - Machine Learning
University of Maryland
08.2021 - 12.2022
Bachelors - Computer Science and Engineering
PES University
08.2015 - 08.2018
Skills
Python
R, MATLAB
C/C, Javascript
Database: SQL - PostgreSQL, NoSQL-MongoDB,DynamoDB
Software Engineering: Linux, Git, Jenkins
Cloud: AWS, Azure, GCP
Data Engineering: PySpark, Hadoop, GlueETL
Machine Learning: AWS Sage Maker, Numpy, Pandas, Scikit-learn
Deep Learning: Tensorflow, Keras, PyTorch, HuggingFace
Computer Vision: OpenCV
Natural Language Processing: Spacy, NLTK
Certification
AWS Machine Learning
Timeline
AWS Machine Learning
03-2023
Deep Learning Specialization
12-2022
Machine Learning Research Assistant
University of Maryland, College Park
02.2022 - 12.2022
Master of Science - Machine Learning
University of Maryland
08.2021 - 12.2022
Machine Learning Engineer
TensorFlow | NLP | PySpark | AWS | ETL, Akamai Technologies
07.2018 - 08.2021
Bachelors - Computer Science and Engineering
PES University
08.2015 - 08.2018