Active Hackathon

Deep Dive: Dunzo Ups Its Machine Learning Game To Reach The Mark Of 25 Million Orders Per Month

Over the past few years, the delivery services in India has witnessed a massive growth. Today, we have a delivery service for every task — whether you want to order food, get something delivered from one point to another in a city etc. However, there was a small void in this sector as well (for every task there is a different app) until Dunzo came into the scenario.


Sign up for your weekly dose of what's up in emerging technology.

Founded in 2014 by Mukund Jha, Kabeer Biswas, Dalvir Suri, and Ankur Agarwal, Dunzo in recent times has gained a significant amount of traction in the industry.  And in order to know more about the industry and how the company operates, we caught up with Mukund Jha, Founder and CTO at Dunzo

Many Needs. One App

Dunzo is an on-demand delivery service platform has changed the way people move things, shop and commute. It does 60-minute deliveries and it serves in three verticals: commute, courier, and commerce.

Commute: In this space, Dunzo provide on-demand bike taxis that people can book through the app and travel from point A to point B. As of now, this service is only available in Noida, Gurgaon, and Hyderabad.

Courier: This option lets a customer delivery something from one part of a city to another. All s/he needs to do is put the pick-up location and drop location on the Dunzo app, and a delivery person will be allocated to do the job.

Commerce: Talking about commerce, this is 70% of Dunzo’s Transaction. People can buy anything from any store, without even stepping out of their homes. All they need to do is set the store location, mention the items they want to buy, and a delivery person will do the job for them.

“We are a full-stack, horizontal service. We do everything that is under the sun,” said Jha. “And being this full-stack fulfilment service, we control the supply, we control the merchants that work with us, and provide us with a deeper sort of understanding of the ecosystem and that allows us to remove friction from multiple steps.”

The Use of AI and ML

Looking at the technology side at Dunzo, the company rely on its mobile app which is available for both Android and iOS. And the infrastructure behind consist of Python, Jango, NodeJS and Go.

That is not all, there is one more thing that plays a crucial role at Dunzo — Data. “At Dunzo, we breath data day in and day out. Every decision that we make has some sort of data backing behind them,” said Jha.

Dunzo has a large analytics setup where every action that a user does get recorded and stored in the company’s private data store. Once the data gathering is done, a team of data science professionals analyse that data and come up with things that can help in making Dunzo services better.

In terms of machine learning and artificial intelligence, Dunzo is making the best out of these sought after techs as well. The company uses machine learning to decide on some of the major aspects such as the personalized content and offers for users, to figuring out the best partner to get a task completed, to figuring out the best merchant to get your order fulfilled etc.

“We are already using machine learning and data, but we want to take it to the next level. And we are making sure that machine learning and data becomes an integral part of everything that we do at Dunzo — right from figuring out on which cities to launch to what categories to focus on how do we really personalize the user experience,” said Jha.

The Partner Allocation Algorithm

One of the most crucial things in on-demand delivery services is task allocation. If you are a Dunzo user you might have noticed, every time you place an order the app allocates a delivery person who fulfils the task for you. But, have you ever wondered how Dunzo does it? What are the parameters that it considers before allocating the best suitable partner for the task?

To select the best partner or the best delivery person, Dunzo uses a complex algorithm that takes a bunch of things into account.

For example, if there is a grocery order, Dunzo’s partner allocation algorithm figures out whether a partner has visited a grocery store before, has he done this kind of task before, how many deliveries the partner has done in that area etc. So, based on these aspects, Dunzo allocates a delivery partner for a task.

Further, there are other parameters as well, such as distance and when they see partners tapering off — Dunzo’s algorithm tries to bring them back on the platform with more relevant and convenient tasks closer to home.

Dunzo’s Way Of Hiring

Being a complete data and technology-based platform, Dunzo makes sure that it hires the best engineering and science talent. The company follows a Google-style hiring procedure. When hiring for the engineering side, it focuses on algorithms, data structure etc. and there is also a machine coding round where a candidate needs to code and solve a problem.

Talking about the data science side, the hiring procedure includes a take-home exercise that the candidate solves with implementing things by taking their own time. Furthermore, Dunzo also makes sure that the candidate has a strong hand on the fundamentals of data science.

“A lot of people are really good with applications, but at Dunzo we also want to have people who are really strong with fundamentals. Because the applications and the problems keep changing,” said  Jha. “So, for example, if someone has previously worked on SVM, we make sure that the candidate has a fundamental knowledge of SVM.”

Coming to the last hiring rounds, the company presents open-ended problems to the candidates and see whether they are able to think through different parameters and whether they are able to solve the problem keeping business in mind.

Also, Dunzo is aggressively hiring for deep learning technologies now as it believes it also going to be an integral part of the company and its operations.

The Roadmap Ahead

Looking into the future, Dunzo is planning to its order fulfilment to a much higher stage. At present, the company completes 1 million order on a month and it is looking to make it 25 million by the end of next year. Furthermore, it is also planning to expand its engineering team from 80 people to 200 people.

“There is a large trend towards on-demand needs — users’ expectations are changing; people want things to be delivered to them fast. And keeping all that in mind, we want to be the best player in this on-demand space,” Jha concluded.

More Great AIM Stories

Harshajit Sarmah
Harshajit is a writer / blogger / vlogger. A passionate music lover whose talents range from dance to video making to cooking. Football runs in his blood. Like literally! He is also a self-proclaimed technician and likes repairing and fixing stuff. When he is not writing or making videos, you can find him reading books/blogs or watching videos that motivate him or teaches him new things.

Our Upcoming Events

Conference, in-person (Bangalore)
Machine Learning Developers Summit (MLDS) 2023
19-20th Jan, 2023

Conference, in-person (Bangalore)
Data Engineering Summit (DES) 2023
27-28th Apr, 2023

Conference, in-person (Bangalore)
MachineCon 2023
23rd Jun, 2023

3 Ways to Join our Community

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Telegram Channel

Discover special offers, top stories, upcoming events, and more.

Subscribe to our newsletter

Get the latest updates from AIM