Industry Research and Users Symposium (IRUS)

Background of IRUS sessions

Industry Research and Users Symposium (IRUS) has always been at the forefront in bringing together solution providers and users of HPC, Cloud and Data platforms. This has been an open and engaging platform where they can share their challenges and successes and discuss relevant technology issues related to computational and data sciences. Previous IRUS sessions (2019, 2018, 2017, 2016) had tremendous participation from conference attendees and included senior speakers from industry like Amazon, Flipkart, Intel, Microsoft, Shell and Xilinx, academia like CERN, IISC and TIFR and start-ups such as Khosla Labs and Forus Health.

2021 IRUS session

This year’s IRUS will continue to bring attention to emerging themes in the areas of computational and data sciences from leading academics and industry practitioners. The sessions have been specially designed to have a diverse topics and speakers that not only matter to deepening our understanding of cutting-edge breakthroughs but also enables how their impact can be widened across industry and society.

Topic
Production Readiness of ML and Data-driven Products and Solutions

Abstract
The exponential growth in AI and ML research is motivating its adoption for enterprise applications at large scale for automation, reducing carbon emissions, real-time contextual recommendations, etc. However, systems performance (throughput and latency) of AI based deployments do not get the same coverage. Apart from high accuracy, performance is also a key concern that needs to be addressed before an AI based application is production ready. The key questions we hope the speakers will address in this forum are as follows.

1. 1. Are AI/ML algorithms generalizable enough for unseen data sets ?
2. 2. Is the system used to deploy these models good enough to support the throughput and desired user latency?
3. 3. Can these models be deployed on resource constrained devices – closer to the user?
4. 4. Can we exploit available heterogeneity (such as CPU, GPU, FPGA, etc.) in the available system for better utilization?
5. 5. How to deal with vocabulary and expertise mismatch between data science and systems engineering to accelerate Al pipelines including pre-processing stages, model deployment architectures, etc?
6. 6. What is the effect of model drift in production to development, MLDevOps?
7. 7. How effective are cloud deployments?
8. 8. Is it desirable to retrain a model frequently with real time data? Do we need a single unified system for operational, analytics and AI workloads ?
9. 9. Is it true that training a model is a one time job and training performance is not so important?
10. 10. Is the cost of infrastructure for serving AI based inferences a big concern?

Session Moderator
Dr. Rekha Singhal

Speakers (in alphabetical order of first names)

Dr. Biswanath Pati an MCA and Ph.D is a senior Data Scientist working in TCS. He comes with over all 17+ years of IT Experience and he has worked in many large engagements both at onsite and offshore for some of the most respectable customers in the IT domain. Since last 3 years he has been working in the Computer vision area for TCS Optumera Recognyze (Plannogram Compliance) product. He won 1st Prize in TCS TACTiCS conference in 2017, Won Ideathon for a large European bank in 2018 and a TCS innovista award for the team in 2019. He is a good story teller and has also authored a fiction titled “One Week With Her”.

Deepa Jayaveer is a chief architect of Optumera, TCS retail AI suite of products. She is a highly accomplished TOGAF certified architect with strong technical knowledge of core retail systems and architecture in multiple areas including machine learning, Big data, Cloud platform, Microservices and Devops automation. She is a proven mentor and has expertise in communicating across strategy initiatives with Product Engineering teams to drive product vision and foster culture of excellence.

Dr. Gaurav Aggarwal is a research scientist at Google India. His interest lies in machine learning and computer vision. Before Google, he worked at Ola Cabs, a ride-hailing company in India, as the head of data science. Earlier, he founded a technology start-up Fashiate that was acquired by Snapdeal, an e-commerce marketplace. He has also worked as a senior research scientist at Yahoo Labs and as an assistant professor at the University of Notre Dame. He did his B. Tech in Computer Science from IIT Madras and PhD from the University of Maryland, College Park.

Mayuresh Jog has 17 years of TCS experience working across multiple global customers in US, Europe and India. He has a wide variety of experience in Big data, Data governance, AWS, ECM and Java technologies. Currently, Mayuresh is working as a SolutionArchitect for a large private bank in India for implementation of their data lake program in Big Data technology.

Neha Malik is a data scientist and BFSI analytics practitioner with 11 years of work experience. She is an electronics engineering graduate and an MBA (Finance). She
has worked with Accenture, State Bank of India and Volkswagen Finance before joining TCS as an Analytics Delivery Lead in BFSI vertical. She is a TCS Contextual Master and has managed delivery of data science & big data ML projects in the area of risk management, liabilities analytics, digital banking and customer analytics.

Dr. Puranjoy Bhattacharya is a Sr. Principal Data Scientist at Infosys. In his current role in the service industry, he works hands-on as a data scientist in the areas of advanced analytics and AI, providing consultation across business verticals, and spanning theentire spectrum of processes starting with customer presentations, providing responsesto RFPs, solution definition and architecture, all the way down to execution and deployment. To the organization, he provides strategies, roadmaps and structures for training.He also drives computer vision in his unit, where their solutions in retail have good traction today. Dr. Puranjoy has over two decades of past experience in R&D roles in industry. His core technical expertise includes data science, deep learning, computer vision, deep learning based NLP, signal processing and pattern recognition for speech, audio and biomedical applications, and S/W design. He has the bent for diving deep into topics and converting research outputs into working solutions, and has a number of papers and patents to my credit. His current focus is in helping organizations to convert their data, whether in structured or unstructured form, into meaningful insights that aid data driven decision-making at every level through models that utilize machine learning and AI in ways that are at once practical, efficient and ethical. He believes that the role of AI is to bring focused information and analyses to the fingertips of decision makers, enabling individuals to perform at enhanced levels and concentrate on what brings maximumvalue to the organization.

Saubhik Baral has 20+ years of experience in business consulting, product engineering and business development being part of the TCS ecosystem. For the last 3+ years, hehas spent his energy in building and enriching multiple AI models in the areas of Source to Pay and Talent Management, acting as head of the analytics team. In addition to that, his team also built up multiple transactional visualizations to arm the business users with real time analytical insights. As per his role, it was critical for him to identify the real time business problems and drive productivity improvement, accuracy, predictions and risk identifications. The AI models as developed, have significantly helped TCS customers to realize the value and trust in application of artificial intelligence in the industry. Saubhik possesses a master’s degree in applied sciences from University College of Science and Technology, Kolkata and also a master’s degree in business management from Indian Institute of Technology, Kharagpur.

Vijayendra (Viju) Shamanna is the Vice President of AI Labs at Ushur. He was previously Ushur’s Senior Director of Engineering and Data Science. He has led all of the company’s key ML innovations over the past three years. Viju brings almost twenty-five years of experience in software and systems engineering, much of it spent in leadership positions spanning domains such as data center infrastructure, cloud-native applications and applied machine learning. Prior to Ushur, Viju bootstrapped the India engineering team for Vexata, a disruptive enterprise storage startup acquired by StorCentric. At Sandisk India, he led the emerging systems group that developed hyper-scale, disaggregated flash storage systems optimized for big data.

IRUS Co-chairs
Laks Raghupathi, Shell India – laks.aarthi@gmail.com
Sundar Dev, Google USA – sundarjdev@gmail.com

Are you interested to organise an IRUS session in the coming years? 
Write to us at info@hipc.org/hipc2021