Apna leverages MongoDB Atlas to connect 30 million job seekers with the hyperlocal job opportunities

INTRODUCTION

Connecting skilled workers to local jobs in 74+ Indian cities

Apna's talent-matching algorithm has enabled successful pairing of suitable candidates with relevant job openings, ultimately bridging the gap between employers and job seekers. As a result, the platform is playing a vital role in promoting employment and fostering economic growth in India.

“India is still developing its economy, and unemployment is rife,” says Suresh Khemka, Head of Platform Engineering at Apna. “People want to work. There are opportunities out there, but they’re hard to find. We provide a central platform for job seekers to find jobs, advertise their services, and connect with their local communities.”

Founded in 2019, Apna is already India’s leading jobs and professional networking platform. It operates across 74 cities and is used by big-name companies, such as Burger King, Zomato, and Delhivery, and by tradespeople, like delivery drivers, service workers, and more. The company also partners with leading public sector institutions, such as UNICEF, YuWaah, and the Ministry of Minority Affairs of India, to support the National Skill Development Corporation.

THE CHALLENGE

Breaking away from the shackles of monolithic architecture

Apna users can access unlimited job listings by creating an account with just their phone number. But with 30 million users and 400,000 employers posting up to 700 times a month, the company needs to do more than simply display information.

Data needs to be relevant, personalized, and highly available to help match the right person to the right job. Apna also needs to verify employers posting on the platform to protect workers from fraud.

“As a start-up, we wanted to get moving quickly, but when the business began to grow we realized our monolithic infrastructure was slowing us down and lacked scalability,” recalls Khemka. “We need to be able to hone our app and roll out new features quickly.”

Apna switched from monolithic to microservices in September 2021 but needed more flexibility than its PostgreSQL relational database, which was integrated with Elasticsearch, offered. Not only difficult to scale, it was also becoming costly and complex to maintain, which led to outages and performance issues.

“We had 100 microservices running and two viable options: find a great database and learn how to run it in house; or use a cloud-based solution,” Khemka explains. “We decided to look for a cloud solution so we could focus on developing our platform.”

THE SOLUTION

A high performing and scalable primary database

Apna migrated to microserves in September 2021 and evaluated various options for data stores. MongoDB Atlas was selected and the company ran training sessions to educate the team, citing the solution’s high reliability as giving them greater confidence that the project would be a success. Apna gradually moved key services out of its monolith environment and started building new features using microservices.

MongoDB Atlas is a reliable, scalable data platform that offers high performance. It’s user friendly, supports auto-scaling, has a flexible data model, and therefore, was the obvious choice for Apna. The platform acts as the primary real-time data source for the content management system. Every interaction that happens on the app is facilitated by MongoDB, from searching for job listings to inviting individuals to apply for roles.

“Creating a user profile was becoming a bottleneck on our old system. MongoDB Atlas can handle 28 million active users without a hitch. That means we can start asking for more data to get users more engaged in the platform and shape better experiences,” says Puneet Kala, Head of Marketplace Engineering at Apna. And with MongoDB Atlas Cluster Auto-Scaling, the company can seamlessly add more data points while using the MongoDB dashboard to keep costs in check.

The team is also building out more functionality for developers and capturing richer data to be used for reporting and analytics. Data moves seamlessly between MongoDB Atlas and a third-party data warehouse for business intelligence.

“When you introduce a new technology there are often teething problems, but we haven’t had a single performance issue. MongoDB Atlas is the leading document database on the market,” explains Ranveer Singh, Head of Engineering, Community at Apna. “I’m confident it can handle anything we throw at it without causing performance issues.”

Apna’s solution architects also get support from MongoDB professional services when they’re developing new solutions, and the more the team can run on one database, the more productive and efficient they can be. Apna has now established best practices to manage its infrastructure at scale and support a variety of use cases.

THE RESULTS

Setting the scene for a more personalized experience

Since implementing MongoDB Atlas and adopting microservices, Apna has seen a 90% reduction in incidents such as failed deployments, changes failing in production that led to outages, and production outages caused by performance degradation as the company’s explosive growth outpaced the capabilities of the monolithic environment.

Developer productivity has also increased by 40% thanks to moving to microservices and adopting a flexible schema model. This helps the team build new features and make enhancements faster, which means more of them can be deployed in a single day. Microservices can also be deployed in isolation, further speeding up development.

With the app functioning well and is highly available to users, the company is free to focus on new initiatives such as how to monetize its data.

“MongoDB Atlas totally takes away the stress from my team. We don’t have to worry about setting up or managing another solution. It’s our de-facto database of choice and first port of call for all new services because it’s so reliable and flexible,” Khemka added. “We have a ‘why anything else’ approach around deploying new technology and the team would have to put forward a strong argument for not using MongoDB Atlas.”

This flexibility is key to developing new features. Apna’s team can iterate quickly, capture user feedback, and continuously optimize the human experience. “We experiment a lot and roll out changes incrementally,” Ravi Singh, Principal Architect at Apna revealed. “Sometimes different regions have different requirements, and now we can cater to those differences and start to build a more personal, relevant experience.”

The team is currently testing MongoDB Atlas Search and looking into more advanced analytics. Apna is a young company, and data is its lifeblood. With the right technology in place, it can explore how to use data to better address India’s unemployment issues – for example, while today it matches workers to opportunities, in the future it could use analytics to understand which skills are in demand and in which regions to help workers make smart choices about the training they invest in.