Pratik Ratadiya > Work Experience
A list of my professional work experiences in reverse chronological order. My academic research experience is listed here. Brands really bring a difference, don't they?
- NASA Jet Propulsion Laboratory Pasadena, CA Visiting Researcher: Mission Control Systems Deep Learning Technologies
- Apple Inc. Seattle, WA Machine Learning Intern: Machine Intelligence - Machine Translation (Automation)
- Johnson & Johnson Pune, India Data Science Consultant: Medical Instruments
- vCreaTek LLC Pune, India Data Scientist: Machine Learning and Product
- Indian Institute of Space Science and Technology Kerala, India Research Intern
- Patagonia Health Pune, India / Raleigh, NC Project Intern
- Indian Institute of Technology, Bombay Mumbai, India FOSSEE Summer Fellow: Spoken Tutorials
- OpenEBS Bangalore, India Project Intern: Mayadata
Mar 2023 - August 2023
Description: Responsible for developing a multimodal anomaly detection system for the power systems of the Mars rovers, published at AIAA SciTech 2024. The developed system was tested on 18 different signal types on over 62K data points. I also designed a natural language interface for this system using the GPT 3.5 API. My second project involved implementing the OpenScene paper for 3D surface reconstruction to make question answering systems on 3D volume renders for Mars surfaces.
June 2022 - Sept 2022
Description: Researched and implemented BERT-based models for offensive and vulgar language detection. Improved the F1 score of existing systems by 15%. Accomplished end-to-end project work: Managing data annotation, training models, deploying REST APIs, and integration with current pipelines.
Jan 2021 - Aug 2021
Description: Staffed by vCreaTek LLC with the responsibility of improving the price forecasting methods of the Medical Instruments Global Division of J&J. Implemented a scalable price forecasting algorithm capable of training 500K+ models for various product-market combinations using a single script. Decreased the MAE at platform-level price predictions by 35%.
Sept 2020 - Sept 2021
Description: Responsible for improving the data science capabilities and software solution consulting of the company. Researched and led a team of 3 people to implement a table-based statement verification model that got a Top 8 worldwide position in the Sem-Tab-Fact task at SemEval 2021. Part of the five-member team at Frontdoor Inc. that developed the first version of their developer portal using Golang. I also redesigned the MVP of Foshan, a CTRM product of the company.
May 2019 - Aug 2019
Description: Researched and developed a multilabel profanity detection system using multi-head self attention ensembles to be installed in the feedback systems for IIST, an academic institute under ISRO. The system achieved a 98.3% weighted F1 micro score on the Kaggle comment toxicity detection challenge.
Sept 2018 - Mar 2019
Description: Worked on exploring and prototyping different machine learning and deep learning models for heart disease detection to be used in the future for the medical records of the company. I also had a chance to demo the work to the founders on their trip to India.
May 2018 - July 2018
Description: Responsible for designing and deploying a multi-level spam filtering systems for the forums of Spoken Tutorials, a project of the Ministry of Education, Govt. of India with over 1.8 million users. I implemented a Naive Bayes and SVM-based solution as a chained ensemble, and later deployed into the existing web application.
Feb 2018 - May 2018
Description: Worked on the existing OpenEBS framework in Go and WeaveScope, a JavaScript based framework used to visualize Kube clusters. Made multiple contributions to their repository. Selected as the Top 5 interns in this program from amongst the 100 interns countrywide.