About

Vishal Burman is a highly skilled Data Scientist with more than 2 years of experience in the field. Currently, he is working at InSyncAI Inc. as a Data Scientist, where he is responsible for conducting research on state-of-the-art (SOTA) techniques in automatic Question-Generation. He has also developed and maintained an end-to-end pipeline for Question-Generation and developed scripts for generating diverse beams to reduce server latency. Vishal has expertise in optimizing models using ONNX runtime and quantization and generating detailed qualitative comparison reports of different architectures. In his previous role as a Software Engineer Intern at InSyncAI, Vishal developed a production-level end-to-end pipeline to generate paraphrases using Transformer. He also conducted research and developed test scripts of SOTA techniques in data-augmentation for text-based classification systems. Additionally, Vishal has developed scripts for text-generation for papers such as "Latent Bag of Words with Gumbel Top-k Reparameterization" [NeurIPS '19] and "Query and Output: Generating Words by Querying Distributed Word Representations for Paraphrase Generation" [NAACL '18]. Before joining InSyncAI, Vishal worked as a Machine Learning Engineer at Wisopt, where he developed and deployed at scale recommender-system based on user click behavior. He improved the existing chat-bot performance by improving the core architecture and conducted research on optimization of ensemble models. Vishal also reduced the latency of ensemble models by optimization using JIT compiler and Cython. Vishal holds a Bachelor of Technology degree in Computer Science from SRM University. He is a Research Scientist, Data Scientist, and Software Engineer with expertise in Bag of Words, ML, and test. With his relevant experience of 2.43 years, Vishal is an excellent problem solver and has a proven track record of delivering high-quality results.