Manish Pandey

Manish Pandey is an adjunct professor in the ECE Department at Carnegie Mellon University where his research interests range broadly across embedded systems, intelligent edge computing, system architecture for machine learning, and natural language processing. Pandey is also a Fellow/VP R&D at Synopsys, where he leads the Machine Learning and Formal Verification engineering teams.

Pandey has extensive experience in machine learning, distributed systems, and infrastructure, and he previously led storage analytics systems at Nutanix and display ad targeting and security groups at Yahoo!. He previously led the development of several formal verification technologies at Verplex and Cadence, which are in widespread use in the industry. Pandey has been the recipient of the IEEE Transaction in CAD Outstanding Author Award and holds more than two dozen patents and refereed publications.

He completed his Ph.D. in computer science from Carnegie Mellon University and a BS in computer science from the Indian Institute of Technology Kharagpur.

Research projects: ServiceNet

The ServiceNet project aims to improve how machines can answer questions directly from large collections of documents. Traditional open domain Q/A systems consist of a pipeline of retriever and machine reading steps that identify the relevant answer span in a set of document retrieved in response to a user query. These are limited in two ways. Retrievers are typically based on statistical word count measures such as tf-idf or bigram hashing variations and have limited semantic matching capabilities. Current state-of-the-art machine reading approaches work at a paragraph-level, and require complex paragraph and document ranking approaches to identify the correct answer, and do not have the capability to directly utilize the entire document for identifying answer spans.

ServiceNet is addressing these challenges using neural information retrieval and pre-trained contextualized embeddings with transformer models. ServiceNet, in addition, uses personalization as well as static and dynamic context to obtain accurate and relevant answers for users.

Current Students:

Past Students:

Talks and presentations

• DAC 2019 Tutorial: Designing Application Specific AI Processors
• ChipEx 2019, Tel Aviv, Keynote: Artificial Intelligence: Driving the Next Generation of Chips and Systems
• DAC 2018, ML Workshop Keynote: Transforming Machine Learning with EDA- Opportunities and Challenges
• NSF SRC Invited Talk, July 2018: Explainable Design Automation with Machine Learning