India, January 5, 2023: Cactus Communications, a science communication and technology company, recently won the ‘Best Innovative Solutions Leveraging Artificial Intelligence (AI)/Machine Learning (ML)’ in the 6th edition of the Amazon Web Service (AWS) AI Conclave. Held on December 14, 2022, Bangalore, this year’s AWS AI conclave deep dived around themes such as accelerating innovation, scaling effortlessly, and creating a real impact for businesses, with AI and ML services on AWS.
Nishchay Shah, Chief Technology Officer, CACTUS, said, “In the field of science communication, we know that researchers have always struggled to publish their papers because of strict deadlines, not being able to readily submit high-quality manuscripts, and language barriers. At CACTUS, our breakthrough product, Paperpal, an AI-powered assistance for language and technical compliance that checks for research manuscripts, was designed to simplify a researcher’s life.
He further added, “The award recognizes our efforts in using AI algorithms and AWS’ suite of products and services that enables us to reach more researchers; automate several of our internal processes; and create new products, leading to significant business growth. We are thankful to our entire team who made this happen and our customers who has always supported us.”
At the conclave, Jalaj Thanaki, Principal Data Scientist, CACTUS, spoke about how CACTUS leveraged Amazon SageMaker distributed training to train the RoBERTa Model, and Nishchay Shah was part of a panel discussion on how AI and ML have shifted from being an aspirational technology to mainstream ones, with automation accelerating efficiency in management, creating a real impact in businesses alongside other key technology leaders from AWS, Paytm, and Capri Global Capital.
The flagship Grammar Error Correction (GEC) model uses large language models and reranking algorithms to run automated editing solutions and suggestions, leveraging grammar error correction and text infilling. This facilitates the real-time delivery of subject-specific language suggestions with fast, superior text. Re-ranking and deduplication algorithms power CACTUS’s personalized research discovery platform, R Discovery, which has 2M+ downloads on app stores and is used by thousands of researchers worldwide. Using the AWS SageMaker Data Parallel API resulted in a 41% reduction in training time with a negligible impact on accuracy.
