Meesho has filed its draft red herring prospectus (DRHP) with the Securities and Exchange Board of India (SEBI) via the confidential route to raise ₹4,250 crore in primary capital through an Initial Public Offering (IPO), reported Moneycontrol on July 4.
The e-commerce platform mainly focuses on customers in Tier 2 and Tier 3 cities across India. The company has raised $1.6 billion (~₹13,280 crores) across fourteen funding rounds, led by investors such as Tiger Global, Think Investments, and Mars Growth Capital, among others. The latest funding round in January 2025 valued Meesho at approximately $4 billion (~₹33,200 crores).
It boasts a total of 187 million annual transacting users, who collectively placed 1.3 billion orders in the nine months ending December 2024, according to a previous company statement. The platform also features over 4 lakh annual transacting sellers across various categories.
Moreover, it became the first horizontal Indian e-commerce company to achieve profitability in FY24 and the first to generate a free cash flow of ₹197 crore for the entire year.
While Meesho’s growth in sales, users, and orders has rightly drawn attention, it has made substantial progress in building the technology that powers its platform.
Beyond being just an e-commerce marketplace, Meesho has developed a robust internal tech stack, powered by AI and machine learning, that supports and drives its operations at scale.
Meesho Open Sources Its Internal ML Platform
“Meesho’s cutting-edge technology operates at an unprecedented scale, making it the only Indian e-commerce platform managing such massive complexity,” the company said. It’s a machine learning platform that processes around 67 trillion feature retrievals (data signals fetched to make real-time predictions) and over 3 trillion inferences (real-time predictions) per day.
Recently, the company also open-sourced several core components of its machine learning platform, BharatMLStack. These include the feature store, control plane, orchestration UI, and various software development kits (SDKs).
“We put BharatMLStack to the test during high-traffic events like our Mega Blockbuster Sale in March 2025, where it delivered at scale, demonstrating its ability to perform under peak load conditions. This helped Meesho drive higher user engagement, better conversions, and increased order volumes during the sale,” Sanjeev Barnwal, founder and CTO of Meesho, said.
“By open-sourcing it, we’re sharing a high-scale, AI stack with the broader tech community—purpose-built for real-time use cases and tailored for Indian businesses,” he added.
The Online Feature Store (OnFS) is a core component of the company’s open-sourced stack. This helps the platform serve pre-computed, up-to-date features instantly for real-time model predictions.
This enabled various use cases like fraud detection and personalised recommendations on the platform with high accuracy and low latency. According to the documentation, the company said the OnFS consistently serves 99% of feature-retrieval requests in under 10 milliseconds.
Moreover, BharatML provides a production-ready machine-learning infrastructure with multi-database backends, including Scylla, Dragonfly, and Redis. The SDK also supports multiple languages, gRPC APIs, and extensive documentation to help engineers and developers leverage the tools on their apps.
The company also outlined all the features, configuration options, benchmarks, and other technical details in the documentation published on GitHub.
Meanwhile, Meesho has also open-sourced a modern MLOps management interface called Trufflebox UI. This provides developers with a modern web interface for managing their entire ML infrastructure.
It offers a comprehensive feature catalogue, discovery, role-based access control and user management, along with a responsive user interface designed for both desktop and mobile devices.
In one sense, Meesho also wants to give back to the open-source community, having largely benefited from it. “We benefit quite a lot from open source. A lot of it (Meesho’s tech stack) is built on top of open-source projects. So, it’s time to give back to the community,” Debdoot Mukherjee, the company’s head of AI, said in an interaction with AIM last year.
The company has over 20 repositories on GitHub and regularly publishes technical blogs that outline various updates and developments in its tech stack.
AI for the Bharat User
In addition to building a high-performance machine learning stack internally, Meesho also offers several user-centric generative AI capabilities on its platform.
The app features large language models (LLMs) that help users search for various queries in the app, using their native language. These LLMs enhance search by decoding various vernacular expressions, correcting errors, and refining queries for precise results.
“Today, we are at a stage where every function, or pillar of the company or marketplace we run, is powered by AI in a significant way,” Mukherjee said in a podcast episode earlier this year.
“Right from providing a very simplified procedure [for sellers] to list products on the platform, to enabling sellers to think about how to price products in the right way for the sales to grow, and how to run ads on the platform efficiently. All of these problems are being powered by some AI model at the end of the day,” he added.
Moreover, the company’s in-house logistics division, called Valmo, which was built with the intention of streamlining the company’s supply chain, also utilises AI to a large extent.
This utilises AI to help the platform accurately identify addresses and geocode them to their corresponding locations, while efficiently allocating orders for fast deliveries and other customer support tasks, Mukherjee explained.
One of the company’s larger AI initiatives is GenAI-driven customer support. The voice-based agent is available in six Indian languages and is said to resolve around 90% of queries at just one-fifth of the original cost. The voice-bot operates around 60,000 calls per day.
“Voice interfaces are particularly powerful for the audience. We call the user a ‘Bharat User’. This user persona has a fundamental problem in using a Latin keyboard in typing a vernacular text,” Mukherjee said, indicating that text input is difficult for users to type out long sentences.
Besides Meesho, there is a long list of startups in India with a strong focus on building an in-house tech stack to power various services on their platforms. This includes some of the most popular names like Zerodha, Cred, Razorpay, Zepto, Flipkart, and others.
However, the current challenge facing India’s tech ecosystem is demonstrating its capability to develop a sovereign yet fully functional AI model under the IndiaAI mission.
Startups such as SarvamAI, SocketAI, Gan.AI, and Gnani.ai are at the forefront of this effort, and their progress within the country’s initiative valued at over ₹10,000 crore will be closely watched.
The post The Tech Company That is Meesho appeared first on Analytics India Magazine.
