Machine learning, Deep learning, deploying models using AWS Sagemaker

AKASH SINDHU

San Jose, CA, *****

LinkedIn: https://www.linkedin.com/in/akashsindhu/

GitHub: https://github.com/akashsindhu

OBJECTIVE

Machine learning enthusiast with an interest in Data Science, Deep learning algorithms, and MLOps. SKILLS

Languages: Python Cloud: Google AI platform, AWS Sagemaker Frameworks and Libraries: Keras, Scikit-learn,

Pandas, NumPy

Algorithms: Machine learning algorithms, Deep learning algorithms (NN, CNN, RNN, LSTM, Autoencoders)

Others: Git, ETL Data visualization: Seaborn, Matplotlib, MS Excel EDUCATION

San Jose State University (Bachelor’s in Computer Engineering) Graduating May 2020 PROFESSIONAL EXPERIENCE

Machine Learning Intern, WING AI TECHNOLOGIES, (getwingapp.com) (Sep 2019 Present)

• Developed a text classification model using pre-trained GloVe word embedding and transfer learning to classify types of service requests which increased accuracy from 30% to 70% with 329 data points.

• Deployed custom model on Google AI Platform and generate predictions on new data.

• Increased efficiency by 3 times and decreased request processing time. Research Assistant, SAN JOSE STATE UNIVERSITY (Feb 2020 Present)

• Comparative Hi-C Analysis for discovery of conserved or diverged regions of genome structure.

• Developing a method to analyze two (or more) Hi-C matrices for discovery of regions in genome in which the two (or more) samples are highly similar (in Hi-C matrix) or highly divergent. Machine Learning intern, SPEEDLEGAL, (Shaforms.com) (Sep 2019 Oct 2019)

• Developed a bag of words model by extracting words from pdf, cleaned them and made the vocabulary of words that represent important information in the documents.

• Implemented a Named Entity Recognition model using pre-trained library spaCy. PROJECTS

AIes an Ios application (Senior year project ongoing)

• Managed team of 4 with Agile methodology. Designed and implemented iOS application that will narrate surroundings in the form of speech to help visually impaired people.

• Developed, trained and deployed object detection custom model on MSCOCO dataset using keras on AWS Sagemaker and integrated it with the app using API gateway and serverless AWS lambda. Stock Market Prediction using LSTM

• Extracted stock market historical data of more than 2000 companies using Alpha vantage API.

• Compared RMSE values and visualize the predicted data to the test data using moving average, linear regression, random forest, and long short-term memory (LSTM). LSTM performed best with RMSE values least among other algorithms.

Lung cancer detection using CNN (ongoing)

• Developed a medical imaging classification website to detect the type of lung diseases and achieved 90% accuracy. Integrated features like payments are supported. Miscellaneous projects

• Multiclass classification of flower species (Accuracy is: 96.67%), Handwritten digit Recognition (MNIST) using CNN in keras (Accuracy: 99.07%), Object recognition in photographs (CIFAR-10) using CNN (Accuracy: 81.31%), Predicted sentiment from IMDB movie reviews (Accuracy: 88.65%).

• Titanic: Machine Learning from Disaster competition from Kaggle (TOP 27% with 3076th rank out of 11403)

• Implemented and deployed text classifier to predict tag of questions from stack overflow dataset using bag of words approach to google ai platform. Interpreted model with SHAP.

Contact this candidate