Dev Mehta Email & Phone Number

Texas, United States

Chicago, Il, Us

● Led a cross-functional team of 5 engineers, overseeing data collection, preprocessing, and frontend integration. ● Fine-tuned state-of-the-art LLM models like OpenAI's Whisper and Seamless, improving WER by 25% for transcription and translation of multilingual audio data. ● Developed and deployed an end-to-end real-time multilingual audio translation system on GCP, reducing processing time from 3 seconds to 0.3 seconds.

● Conducted target market studies, analyzing audience distribution by zip and city, providing geographic ad reach insights. ● Boosted client confidence by 15% via rigorous ground truth studies, enhancing ad insights by aligning impressions with actual attendance. ● Built an automated data collection pipeline to provide crucial audience demographics, increasing efficiency by 40%. ● Created dashboards for audience demographics, including average distribution, affinity, and in-market/out-of-market breakdown.

● Design course curriculum, create instructional materials, and instruct/assess up to 100 students in diverse classroom environments.

India

● Enhanced Lung Cancer Detection accuracy from 84% to 94% by implementing a YOLOv3-inspired model architecture.

India

● Created a sentiment analysis system for a theater company, utilizing scikit-learn to train diverse machine learning models on their dataset, enabling accurate classification into five distinct sentiment classes ranging from negative to positive.● Designed an English-to-French translation system for a travel company, incorporating a seq2seq approach for sentence tokenization and implementing an LSTM model with Tensorflow and Keras.● Developed a Traffic Sign Classification System using Tensorflow and Keras for a driving institute, achieving 95% accuracy bolstering traffic sign recognition capabilities for driver training.● Created an Image Noise Reducer employing auto-encoders, elevating classification accuracy from 26% to 84% for a security firm.

India

● Employed CountVectorizer and cosine similarity to construct a movie recommendation system, conducting comprehensive movie similarity assessment via genre and keyword-based feature analysis for a movie ticketing company.● Conducted preprocessing using NLTK and parsed resumes using ResumeParser to extract pertinent skills for a job board.● Developed a Flight Delay Prediction model for a tourism company using a Decision Tree Classifier, achieving an accuracy of 88%.

Master Of Science - Ms, Computer Science, 3.88/4.0

Courses: -Introduction to Machine Learning -Introduction to Data Science -Natural Language Processing -Advanced Machine Learning -Machine.

Bachelor'S Degree, Computer Science, 7.48/10

High School Diploma (12Th Grade), Computer Science, 87.4%

10Th Grade, 9.60/10