NASSCOM Annual Technology Conference 2019

Topic: Natural Language Processing (NLP) using Deep LearningBy: MakeMyTripPresenter: Madhu Gopinathan, Senior Vice President, Data Science, MakeMyTripTime: 9:00:am – 1:00 pmVenue: Pearl Ballroom IIPresenter Bio: https://www.linkedin.com/in/madhugopinathan/Deep Dive Brief: Deep learning models have made impressive advances in recent years on natural language processing tasks. At Google, a deep learning based machine translation system reduces translation errors by an average of 60% compared to the earlier phrase-based production system . This improvement in accuracy has been achieved with almost a 100x reduction in code for feature engineering. In this session, we will focus on the foundational concepts behind these models: distributed representation, convolutional & recurrent neural networks, sequence to sequence models and attention mechanism. We will also discuss how to apply these concepts to solve real world problems such as Myra, MakeMyTrip’s customer care chatbot which uses deep learning models for intent classification / slot extraction in production.
This session will take you from the basic concepts of natural language processing to latest modelling techniques using deep learning. We will discuss core concepts and also discuss their implementation using Python: for those who want to follow along, please bring a laptop with internet access to reach http://colab.research.google.com/.
References:

• [1] A neural network for machine translation, at production scale (https://ai.googleblog.com/2016/09/a-neural-network-for-machine.html)
• [2] Google’s Neural Machine Translation System: Bridging the gap between human and machine translation (https://arxiv.org/abs/1609.08144 )
• [3] https://twitter.com/DynamicWebPaige/status/915326707107844097
• [4] Advanced Dialog Systems (http://web.stanford.edu/~jurafsky/slp3/25.pdf )

Broad Level Agenda:

1. Vector space models for web search
2. Word embeddings
3. Neural language models
4. Attention
5. Pretrained language models
6. Applications

Key Takeaways from the Deep Dive:

1. Understand basic concepts in NLP
2. Understand major recent advances in NLP based on deep learning methods
3. Get a sense of the kind of applications that can be powered by these recent advances

Deep Dive Pre-requisites (Any reading or reference material that the participants need to refer to before attending the session)http://ruder.io/a-review-of-the-recent-history-of-nlp/
Hardware Requirements: (Any specific hardware requirements for the session)
If you want to do hands on exercises, then please bring a laptop with internet access to reachhttp://colab.research.google.com/.

Topic: Building First Machine Learning ModelBy: Publicis SapientPresenter: Punit Mishra and Varun SharmaPunit Mishra is currently working as a Vice President Technology. He has been associated with Publicis Sapient for last 18 years and has played various roles starting as a developer to a solutions architect, designing and delivering complex trading and data platforms for asset managers and delivery lead for many of Sapient's financial services clients. Punit is an active member of I&D Committee of Publicis Sapient India and also a lead of women in technology (WIT) track our women leadership network
Varun Sharma is a Senior Manager at Publicis Sapient and is responsible for delivering large-scale complex data driven programs for financial clients spread across multiple geographies. He has 15+ years of extensive experience in delivering complex programs with ever-evolving AI, Machine Learning and Deep Learning technologies and frameworks.
Deep Dive Brief:AI, Machine Learning and Deep Learning is everywhere. Machine learning is so universal today that you probably use it dozens of times a day without even knowing it. Many researchers also think it is the best way to make progress towards human-level AI. These technologies brings together computer science and statistics to harness the predictive power. It is a must-have skill for all aspiring data analysts and data scientists, or anyone else who wants to use all the raw data into refined trends and predictions.
In this session, you will go through the end-to-end process of investigating data through a machine-learning lens. We will walkthrough on how to extract and identify useful features that best represent your data, a few of the most important machine learning and deep learning algorithms, and how to evaluate the performance of your model. More importantly, you'll learn about not only the theoretical underpinnings of learning, but also gain the practical know-how needed to quickly and powerfully apply these techniques to your current business problems.
Machine learning capabilities in Cloud will also be showcased, along with the advances in the field of Automated Machine Learning.
Broad Level Agenda:

1. Introduction to AI, Machine Learning and Deep Learning – [20 mins]
2. Regression, Classification and Clustering techniques – [20 mins]
3. Deep learning [30 mins]
4. Feature extraction, Dimensionality Reduction, Model Selection & Boosting – [30 mins]
5. Build your first Machine Learning and Deep learning model in R & Python – [60 mins]
6. Automated Machine Learning – [30 mins]
7. Machine Learning in Cloud – [20 mins]

Key Takeaways from the Deep Dive:Participants will take back good fundamental concepts of Machine Learning using different libraries in R and Python.
Beginners to Machine learning will learn how to develop and validate their first model and intermediates will learn about how to leverage Auto ML to develop model for complex use cases.
Cloud integration will also demoed.
Deep Dive Pre-requisites

• Introduction to Machine Learning: https://www.youtube.com/watch?reload=9&v=IpGxLWOIZy4
• Introduction to R & Python: https://www.youtube.com/watch?v=6LWnvC2_DZI

Hardware Requirements:Wifi, A/V Support and Projector

Topic : Anthos, bringing hybrid and multi-cloud to enterprisesBy : Google CloudPresenters : Amruta Joshi, Head of Strategic Cloud Engineering, APAC, Google CloudAmruta leads Google’s Strategic Cloud Engineering efforts for the Asia-Pacific region. Her team is responsible for architecting, developing, and deploying Google’s Cloud products for most strategic clients in the region.
Prior to joining Google Cloud, Amruta has worked in various leadership roles in product management, partnerships, research, and software development. Amruta has a Masters in Computer Science from Stanford University and a PhD from UCLA. She serves as a program committee member and reviewer at top tier international computer science conferences including ACM VLDB, ACM WWW, IEEE ICDE, and ACM SIGKDD. She is often invited to give technical talks and keynotes at prestigious venues and international conferences.
Suppprting Speaker:Suchit Puri: Strategic Cloud Engineer, Google Cloud Suchit Puri is a Strategic Cloud Engineer at Google Cloud. He is passionate about DevOps, Networking in Cloud and effective managing your virtual Infrastructure.
Prior to Google, Suchit co founded Adalyz, a cross channel advertising analytics platform. He has worked at Wingify, ThoughtWorks, Progress Software in the past. Suchit holds a masters degree in software systems from BITS Pilani.
Supporting Speaker:Kumar Saurabh: Strategic Cloud Engineer, Google Cloud, Kumar is a Strategic Cloud Engineer and has been Google for about 2.5 years now. He focuses on making organizations successful using Machine Learning and Data Analytics on the Google Cloud Platform.
Prior to joining Google, Kumar has worked with companies like the Royal Bank of Scotland and HT Media in a 5 years long career. He has been involved with developing of a host of Machine Learning solutions for various use cases across these companies. Kumar holds a Bachelor’s Degree from IIT Bombay.
Deep Dive Brief :Anthos, Google Cloud’s new open platform that lets you run an app anywhere—simply, flexibly and securely. Embracing open standards, Anthos lets you run your applications, unmodified, on existing on-prem hardware investments or in the public cloud.
In this session you will learn about how Google Cloud can help you move to a Hybrid Cloud Deployment while continuing to use your existing on-premise investments. Using Anthos built on Kubernetes and Istio we will show you how to bring the efficiency, speed and scale of the cloud to one's Data Center. No matter if you are looking to modernize in-place, or by moving to the cloud, we’ll show you how these tools and technologies can help you build reliable, secure, and high-performing cloud services for today and the future.
Broad Level Agenda:09:00 - 09:10Introductions09:10 - 09:30Google Cloud Platform Overview09:30 - 10:00Hybrid / Multicloud10:00 - 10:30Hybrid / Multiclou10:45 - 11:15Kubernetes Overview + Deepdive11:30 - 12:00Kubernetes Demo12:00 - 12:30Anthos Overview + Deepdive12:30 - 13:00Anthos Demo12:45 - 13:00Wrap up and Q&AKey Takeaways from the Deep Dive:

• Understand hybrid deployment scenarios in Google Cloud Platform.
• Learn about Anthos.
• Learn about Google Kubernetes Engine.

Deep Dive Pre-requisites : Internet to run the demosHardware Requirements: Connectors for laptop (Mac and Chromebook)

Topic: Become Kubernetes Expert - From Zero to Hero in 4 hoursBy: XebiaPresenter: Shekhar GulatiPresenter BioShekhar Gulati is Chief Technology Evangelist at Xebia with extensive experience in Microservices, API first development, Containers, and Serverless technologies. Prior to Xebia, he has worked as technology evangelist for Red Hat talking about OpenShift at conferences around the world. He is an active blogger and has authored a couple of books on OpenShift and JUnit 5.
Akshay Mathur :  https://www.linkedin.com/in/akshmathur/Akshay Mathur is a software architect with 13 years of product engineering experience. His currents interests are Kubernetes and service mesh.
Ravindra Singh Shekhawat : https://www.linkedin.com/in/ravindra-singh-shekhawat-075819108/Ravindra is mission-driven full stack developer and entrepreneur with a passion for Microservices architecture, DevOps practices and virtual reality. He has contributed in Istio service mesh project.
Deep Dive Brief:Kubernetes has become defacto standard for hybrid cloud portable application architecture. With more and more organisations adopting Microservices architecture Kubernetes is the ideal platform for deploying, running, and managing the workload. In this hands-on workshop, we will teach you all the Kubernetes concepts required to run and manage applications at scale. This session will cover:
Broad Level Agenda:

1. Why we are here? Microservices and Cloud Native Architecture
2. The need for containers
3. Installation and getting started with Kubernetes
4. Kubernetes Primitives
5. Building & Running Container Images
6. Environment variables and ConfigMaps
7. Getting inside containers
8. Service to service discovery & load-balancing
9. Rolling updates plus advanced deployment strategies like blue/green, canary
10. Liveness and readiness probe
11. Debugging containers in prod
12. The need for Istio service mesh

Key Takeaways from the Deep Dive:At the end of the session, participants will be comfortable with Docker containers and Kubernetes and can use them at their work place.
Deep Dive Pre-requisites

• Attendees should be comfortable with web application development
• Attendees should be comfortable with a Linux command line

Hardware Requirements: Machine with minikube installed

Topic: Fine Tuning Machine Learning and Deep Learning ModelsBy: MathLogicPresenters: Anurag Verma & Manu ChandraTime: 2:00 pm -6:00 pmVenue: Pearl Ballroom IAnurag is CEO and co-founder FN MathLogic Consulting Services. He has over 23 years of experience in analytics and consulting across North America, Europe, Australia, Africa and Asia. He specialises in using cutting-edge data science tools and techniques for real-world problems. He is experienced in building and leading specialised data science teams in the companies. He has designed and delivered solutions around fraud, credit risk, sales and marketing as well as operational efficiencies with embedded machine learning and deep learning algorithms. Before starting MathLogic he had worked with Accenture, American Express, Genpact and PwC. Anurag holds a Master's degree in Business Administration from IIM, Kolkata and a Bachelor's degree in Engineering from IIT, Kanpur.
Manu is Chief Data Scientist and co-founder FN MathLogic Consulting Services. He has over 18 years of experience in Analytics & Consulting working with leading global organizations in the areas of Fraud Management, Credit risk and Customer analytics. Manu has a passion for teaching and Conducts training in area Machine Learning and Deep learning for leading firms in India as well as Globally. Manu holds a Master’s degree in Business Administration from MDI Gurgaon and a Bachelor’s degree in Engineering from IIT Delhi.
Deep Dive Brief: Practical tips on developing and fine tuning Machine learning and Deep learning models.
Broad Level Agenda: The session would cover the following aspects:

• Introduction to AI/ ML/ DL
• ML/ DL Model performance metrics
• Choice of Error functions
• Choice of Optimization method
• Validation options
• Hyper parameter tuning & Avoiding overfitting
• Data bias in ML/ DL
• Model tuning: RF /GBM/ Deep Learning models.
• Case Studies

Key Takeaways from the Deep Dive:

• High level understanding of AI/ ML/ DL
• Key aspects of building a good ML/ DL model
• Deep dive into model fine tuning process

Deep Dive Pre-requisites:Understanding of predictive modelling process/ machine learning would help participants get more out of the session. Participants may also read first 10 pages from Deep Learning Book. http://www.deeplearningbook.org/contents/intro.html
Hardware Requirements :None. We will be demonstrating on our server. Participants can also follow along on our server as long as they have internet connectivity and any modern browser.

Topic: Modern database for New-age ApplicationsBy: MongoDBPresenter: Vivek Singh, Sr. Solutions Architect, MongoDB,Time: 2:00 pm – 6:00 pmVenue: Pearl Ballroom IIPresenter Bio: https://www.linkedin.com/in/vivek-singh-8b186110/Deep Dive Brief:New applications, users and inputs demand new types of data, like unstructured, semi-structured and polymorphic data. Adopting NoSQL databases means adopting to a new, document-based data model. While most developers have internalized the rules of thumb for designing schemas for relational databases, these rules don't apply to NoSQL. Documents can represent rich data structures, providing lots of viable alternatives to the standard, normalized, relational model. In this session we will cover , why MongoDB is the best way to work with data for developers and how you can take advantage of MongoDB's document model to build modern applications.
Broad Level Agenda:This session will cover MongoDB database feature and its differentiators compared to relational database. Here are the key points covered as part of the deep dive session:

1. Why MongoDB is best way to work with data for developer.
2. How to deliver globally resilient platform through sophisticated replication and failover technology of MongoDB
3. Running MongoDB at scale – sharded cluster in MongoDB
4. Running operational and analytical workloads in the same cluster
5. Build reactive, real-time services using Change Stream
6.  Power of MongoDB Aggregation framework

Key Takeaways from the Deep Dive:(What are participants expected to learn from the session) Benefits of adopting NoSQL database for development & deployment of Modern Applications
Deep Dive Pre-requisites: https://www.mongodb.com/collateral/mongodb-architecture-guide
Hardware Requirements : (Any specific hardware requirements for the session) : None

Topic: Code-mixed Language: Data and SystemsBy: MicrosoftPresenters: Kalika Bali, Monojit Choudhury & Sunayana SitaramMultilingual communities exhibit code-mixing, that is, mixing of two or more socially stable languages in a single conversation, sometimes even in a single utterance. This phenomenon has been widely studied by linguists and interaction scientists in the spoken language of such communities. However, with the prevalence of social media and other informal interactive platforms, code-switching is now also ubiquitously observed in user-generated text. As multilingual communities are more the norm from a global perspective, it becomes essential that code-switched text and speech are adequately handled by language technologies and NUIs.
Code-mixing is extremely prevalent in all multilingual societies. Current studies have shown that as much as 20% of user generated content from some geographies, like South Asia, parts of Europe, and Singapore, are code-mixed. Thus, it is very important to handle code-mixed content as a part of NLP systems and applications for these geographies.
In the past 5 years, there has been an active interest in computational models for code-mixing with a substantive research outcome in terms of publications, datasets and systems. However, it is not easy to find a single point of access for a complete and coherent overview of the area. This session is expecting to fill this gap and provide a foundation in both linguistic and computational aspects of code-mixing. We hope that this then becomes a starting point for those who wish to pursue research, design, development and deployment of code-mixed language technology systems in multilingual societies.

https://www.linkedin.com/in/sugoyal/?originalSubdomain=in

https://www.linkedin.com/in/gargi-banerjee-dasgupta-4509a61/?originalSubdomain=in

https://www.linkedin.com/in/dcternes/?originalSubdomain=hk

https://www.microsoft.com/en-us/research/people/padmanab/

https://www.linkedin.com/in/viralbshah/

https://www.linkedin.com/in/sunilgoyal79/?originalSubdomain=in

https://www.linkedin.com/in/rakshit-daga-76b467/?originalSubdomain=in

https://www.linkedin.com/in/anoop-chaturvedi-629b657/?originalSubdomain=in

https://www.linkedin.com/in/anuragsahay/?originalSubdomain=in

https://www.linkedin.com/in/manojmadhusudanan/?originalSubdomain=in

https://www.linkedin.com/in/arioli-arumugam-333310149/

https://www.linkedin.com/in/nagesh-shenoy-28b2b4a5/?originalSubdomain=in

https://www.linkedin.com/in/manoj-kapoor-0088672/?originalSubdomain=in

https://www.linkedin.com/in/bala-bharadvaj-0400871/?originalSubdomain=in

https://www.linkedin.com/in/jayavaradhansambedu/?originalSubdomain=in