Megh Patel Email & Phone Number

New York, Ny, Us

Scottsdale, Arizona, United States

• Leveraged Google Cloud Vertex AI to implement and fine-tune Google's PaLM 2 (text-bison) for sentiment and text classification of University courses and professors based on diverse review sources• Enhanced and elevated the F1 scores of Text Classification and Sentiment Analysis by 7.7% and 10.8% respectively, transitioning from a baseline BERT model (PyTorch) using LLM prompt engineering, fine-tuning, and few-shot learning• Developed multiple predictive models in the digital marketing… Show more • Leveraged Google Cloud Vertex AI to implement and fine-tune Google's PaLM 2 (text-bison) for sentiment and text classification of University courses and professors based on diverse review sources• Enhanced and elevated the F1 scores of Text Classification and Sentiment Analysis by 7.7% and 10.8% respectively, transitioning from a baseline BERT model (PyTorch) using LLM prompt engineering, fine-tuning, and few-shot learning• Developed multiple predictive models in the digital marketing realm, leveraging Google Analytics data on BigQuery (~3k lines of code) to create features for student enrollment forecasting. Utilized Python and PyTorch within Google Cloud Vertex AI, deploying algorithms such as ANN, Catboost, XGBoost, and Random Forest• Deployed multiple models to production in GCP and created Model Explainability dashboards and Model Monitoring and Drift Detection pipelines Show less

Scottsdale, Arizona, United States

• Extracted nested Google Analytics data using BigQuery and combined with data from multiple sources to create dashboards and features for a Machine Learning model using advanced SQL techniques like CTEs and more• Implemented a Data Drift detection pipeline along with Tableau Dashboard and scheduled Vertex AI managed notebooks for alerts on deployed models using statistical models.• Updated the features to train the model to include the most recent data, retrained the model with the new… Show more • Extracted nested Google Analytics data using BigQuery and combined with data from multiple sources to create dashboards and features for a Machine Learning model using advanced SQL techniques like CTEs and more• Implemented a Data Drift detection pipeline along with Tableau Dashboard and scheduled Vertex AI managed notebooks for alerts on deployed models using statistical models.• Updated the features to train the model to include the most recent data, retrained the model with the new features, and improved the model's F1 score by 5%• Extensive experience in Text Analytics, generating data visualizations using R, Python and creating dashboards using tools like Tableau, Google Data Studio/Looker Studio• Conducted studies, rapid plots and using advance data mining and statistical modelling techniques to build a solution that optimize the quality and performance of data Show less

Tempe, Arizona, United States

• Developed and deployed a scalable Java-based microservice on AWS Lambda using Spring Cloud Function, integrating with AWS SQS for efficient message processing.• Deployed AWS resources using the advanced AWS CDK, optimizing infrastructure with AWS CloudFormation• Designed the service architecture and created integration tests to ensure system reliability and compatibility

• Developed and deployed a robust real-time Computer Vision application for accurate people counting, utilizing TensorFlow and Tkinter on an Nvidia Jetson Nano. Achieved a high accuracy rate of 95% in person detection at a processing speed of 25 FPS.• Implemented advanced object detection algorithms along with object tracking, including DetectNet, YOLOv3, Mask R-CNN, and SSD.• Utilized Transfer Learning and retrained DetectNet Model using Nvidia’s DeepStream SDK on custom dataset •… Show more • Developed and deployed a robust real-time Computer Vision application for accurate people counting, utilizing TensorFlow and Tkinter on an Nvidia Jetson Nano. Achieved a high accuracy rate of 95% in person detection at a processing speed of 25 FPS.• Implemented advanced object detection algorithms along with object tracking, including DetectNet, YOLOv3, Mask R-CNN, and SSD.• Utilized Transfer Learning and retrained DetectNet Model using Nvidia’s DeepStream SDK on custom dataset • Engineered real-time image processing and inference solutions on edge devices using NVIDIA Jetson and TensorFlow Lite, reducing latency by 40%.• Collaborated with data engineering teams to design and implement robust data annotation workflows using Labelbox, enhancing data quality and model training efficiency Show less

Ahmedabad, Gujarat, India

• Designed and Deployed a Food Ordering Chatbot leveraging NodeJS, Firebase, and Google Cloud Platform • Implemented NLP Models such as BERT, ELMO, Word2Vec, and Universal Sentence Encoders with TensorFlow and Transformers (Hugging-face library) to achieve high-accuracy clustering of texts• Optimized chatbot response time by transitioning to a Distributed Database, achieving a reduced webhook processing time with results delivered within 3-4 seconds of the user's request using Parallel… Show more • Designed and Deployed a Food Ordering Chatbot leveraging NodeJS, Firebase, and Google Cloud Platform • Implemented NLP Models such as BERT, ELMO, Word2Vec, and Universal Sentence Encoders with TensorFlow and Transformers (Hugging-face library) to achieve high-accuracy clustering of texts• Optimized chatbot response time by transitioning to a Distributed Database, achieving a reduced webhook processing time with results delivered within 3-4 seconds of the user's request using Parallel Joins and Sort, enhancing user experience• Scrapped data and extracted information using BeautifulSoup, and Selenium libraries in Python• Using Python libraries like NLTK, spaCy, textblob, Gensim, Matplotlib, and numpy and Tableau created Dashboards for sentiment analysis, anomaly detection and other text classification tasks. Show less

Ahmedabad Area, India

Master'S Degree, Computer Science

Bachelor Of Technology - Btech, Information Technology

Education record