We want to use AI to bring in cognitive computing capabilities: Dr Rajesh Narang, CTO, GeM
Government e Marketplace is the National Public Procurement Portal offering an online, end-to-end solution for procurement of common use goods and services by government departments. It is used by Central and State Government Ministries and Departments, Central and State Public Sector Undertakings (CPSUs and SPSUs), autonomous institutions and local bodies. In just two years of its operations, GeM has recorded till date more than 8 lakh transactions worth over Rs 12,239 crore. GeM has nearly 27.9 thousand buyer organisations registered and about 1.43 lakh sellers and service providers offering more than 4.86 lakh products and services for online purchase. Today, GeM contributes to approximately 20-25 per cent average overall savings to the government. As a national platform, scalability and stability of the platform is absolutely critical.
As part of the Indian Tech Leaders Series, we are honoured to bring you the views of Dr Rajesh Narang, CTO, GeM, who shares his views on the technological capability of the platform, the role of open source, and his future vision for GeM
Reproduced below are edited excerpts of the interview:
As the CTO / key architect of GeM, what are some of the key aspects that you have kept in mind while designing the platform for scalability?
Government e-Marketplace (GeM) was launched in POC mode, but soon load on it started growing fast. While launching a fully fledged version of GeM Platform several key aspects have been kept in mind to build design of GeM platform to achieve scalability.
First, an emerging and next generation architecture style known as Micro Services Architecture is used in GeM. As business functionality of GeM can be decomposed into independent business domains so it is fully amenable to the design pattern supported by this architecture style. Each domain is assigned to independent team that has freedom to choose any technology, tool and hardware. Different business domains (modules) have their own codebase, business logic, database and hardware.Several copies of business logic in several servers are deployed horizontally, giving GeM scale ability along X-axis. By decomposing business functionality into Micro Services provides GeM scalability along Y-Axis, and partitioning data of each Micro Service of each business domain on an attribute gives scalability to it along Z-Axis. Thus scalability along all axes is achieved by choosing Micro Services architecture design for GeM.
Secondly, effort was made to get the infrastructure components such as Router, Firewall, bandwidth, Virtual machines, High Security Module and Load Balancers from Cloud Service Providers in virtual mod. Cloud Service Provider was insisted that not only Virtual machines be auto-scalable in horizontal direction but also the other infrastructure components should also scale dynamically on demand as far as possible so that GeM services can be delivered without any disruption.
Thirdly, software such as Docker and Docker Swarm have been used, as they have the capabilitiies to generate new instances if the load on any Micro Service increases to deliver and maintain performance and throughput at desired level.
Four, an innovative Ingress and Egress API design pattern which is an extension of API Gateway Design pattern is used. It allows GeM to collaborate, consume and aggregate the services of other Government departments. By off loading some of the business functionality required in GeM to other departments and orchestrate it in business processes of GeM has allowed GeM to scale out and give good performance to its users.
Fifth, GeM leverages two different Clouds, one Cloud for hosting online procurement transaction processing environment and another Cloud for hosting data warehouse where online analytical processing and reporting are done. This multi cloud architecture gives GeM an unmatched and unparallel Scale to expand. The design pattern of data analytics named as Segregate Write-Read Architecture (SWRA) is a new architecture style which gives near real time data analytics capabilities to GeM.
How important is open source to the overall development and design of GeM?
Open Source software is very important for the overall development and design of GeM since they help in bringing down the Total Cost of Ownership, avoid vendor-locking and allow GeM to share its entire code with community without worrying about complex licensing policy of software. Moreover, policy of Ministry of Electronics and Information Technology (MeitY), Govt. of India promotes the use of Open Source software, so GeM is complying to it. But wherever, it is essential, proprietary software, technology and tools have been used in GeM, for example, they have been used in Data Analytics, Cyber Security and Application Performance Management.
In your view, what are some of the major areas/functions on which GeM has made a major impact in the country? Can you cite some examples?
GeM has transformed the complete purchase to payment process of public procurement for both goods (products) and services which are done in the country. It is an end to end paperless, cashless and contactless electronic system right from registration to payment. It is perhaps the only Portal which offers both goods and services at one place. The details of both Sellers and buyers are verified online from other government databases such as Aadhaar, Income Tax, MCA-21 and GSTN. It has brought transparency, efficiency and speed in public procurement (examples of it are given below) and facilitates buyers to achieve the best value for their money through three modes of procurement-Direct Purchase, e-bidding and reverse e-auction.
GeM supports small sellers since it recommends buyers to take minimum or no EMD for smaller transactions during Bid process so that small sellers can easily participate in several bids simultaneously without locking their capital in different bids. Alternative methods such as bringing down Seller Rating, Suspending Sellers, etc. are adopted to deter sellers from indulging in unethical practices.
Purchases can only be done provided the item pre-exists in GeM sourced catalogue and an item in GeM catalogue is included if no seller has any objection about any of its technical parameters. All the technical parameters are published for wider consultations with all Sellers. It helps in bringing transparency in technical specifications of goods and services. The last procurement prices of the items are shown to Buyers when they put the selected items in Cart to bring in transparency of price and give buyers idea about reasonableness of prices offered on GeM Portal. Sellers have to offer 10% discount in the MRP of the product and sellers are free to offer higher discount on GeM. The price of Product put in Cart is frozen for 10 days, if the price goes up Seller has to offer it to Buyer at the price when it was put in Cart. But if the price goes down, seller has to give product at reduced price.
The overall speed of public procurement has increased with the advent of GeM National Portal. For example, the average closure time from Bid Publication to Payment has been reduced from 25 days in March 2017 to 13 Days in March 2018. It also shows that buyers are more efficient at the closing of financial year compared to other months of the year.
Similarly, the average time taken from Receipt of Goods to Consignee Receipt and Acceptance Certificate (CRAC) generation also came down from 20 days in March 2017 to 4 days in March 2018. Similarly the average time taken from CRAC generation to Payment transfer also came down from 21 days in March 2017 to 5 days in March 2018.
Your future vision for GeM
The vision of GeM is to bring evolution in public procurement by promoting a transparent, efficient and inclusive marketplace. The vision can be achieved by instituting a unified procurement policy that encourages behavioral change and drives reform. Further it requires building an easy to use, fully automated platform which ensures transparency, efficiency in public procurement and delivers new features for inclusive growth and development of all its stakeholders.
DevOps methodology and Artificial Intelligence have been adopted to accelerate the velocity of development and delivery. In fact, in key phases of GeM software project life cycle such as development, build, testing and releases innovative architecture style and tools are being applied to automate them. Further, multiple testing and release environments are being created to design multiple DevOps pipelines to increase overall throughput and quality of features offered in GeM.
The vision of GeM is also to use Artificial Intelligence to bring in cognitive computing capabilities to automate human decisions making process. GeM is already using Live Chat Bot, aim is to use AI Assistants for automating repetitive tasks requiring human judgement, Voice and Image recognition techniques allowing GeM users to search products and services just by voicing over and with the help of images of products. The ultimate aim is to establish a lean, dynamic organization capable of doing continuous innovation, making market driven decisions and committed to delivering value by ensuring right quality at right price.
If you have an interesting article / experience / case study to share, please get in touch with us at [email protected]