Zomato 2026 AI Engineering: Recommender Systems and Entry-Level Roles

Eternal Limited’s central AI team grew from 3 engineers to more than 20 in a single year, and the company signed an OpenAI partnership in February 2026 to build what it calls India’s first AI-native commerce ecosystem.

If you’re eyeing product-company AI roles, Eternal’s engineering surface is worth understanding. The company runs four distinct brands: Zomato food delivery (500+ cities), Blinkit quick commerce, District events ticketing, and Hyperpure B2B supply. Each brand generates ML problems that textbooks don’t cover.

What Eternal Limited’s AI Team Actually Builds

The food delivery company most engineering students in India know as Zomato is now legally Eternal Limited. Regulatory approval for the name change came on March 20, 2025, consolidating Zomato, Blinkit, District, and Hyperpure under a single parent entity. The consumer brands Zomato and Blinkit remain unchanged in the market.

The AI infrastructure underneath those brands changed materially in the following year. In February 2026, Eternal signed a strategic partnership with OpenAI to deepen AI integration across all four brands. According to Business Standard, the scope extends beyond consumer-facing features: it covers internal systems, partner-facing platforms, and Nugget, an AI-native startup being built as part of the partnership.

The central AI team itself is the engineering unit doing this work. Per Times of India reporting, that team grew from 3 engineers to more than 20 within 2024-2025. Their focus: LLM optimisation, fine-tuning, and reinforcement learning across Zomato and Blinkit product lines. Primary stack languages are Java, Go, and Python. Both the Gurugram HQ and the Bangalore campus actively hire.

The ML Problem Space: Recommender Systems, ETA, and Demand Forecasting

Here’s what makes Eternal’s engineering surface unusual: the ML problems span logistics, personalisation, fraud, and supply chain simultaneously. A data scientist at Blinkit might move between demand forecasting and fraud detection in the same sprint. An ML engineer at Zomato might work on restaurant ranking algorithms one quarter and delivery ETA models the next.

The main ML surfaces, based on publicly described role scope:

• Restaurant and dish ranking — personalised restaurant discovery is fundamentally a ranking problem. Collaborative filtering (what users with similar tastes ordered) and content-based filtering (restaurant attributes, cuisine type, pricing) both feed into the ranking model. At Zomato’s scale, this also involves real-time contextual signals: time of day, weather, and previous session behaviour.
• ETA prediction — delivery time estimates require combining traffic data, restaurant preparation time estimates, and delivery agent availability in real time. Getting ETA wrong by even a few minutes affects both user satisfaction scores and order cancellation rates.
• Blinkit hyperlocal demand forecasting — Blinkit’s 10-minute quick commerce model depends on accurate per-dark-store inventory positioning. Too little stock means missed orders; too much means spoilage. The demand forecasting problem at each dark store location is hyperlocal, seasonal, and influenced by local events, making it a harder forecasting problem than typical e-commerce at a regional warehouse level.
• Fraud detection — real-time transaction fraud and fake review detection across both platforms.
• Customer lifetime value (CLV) modelling — predicting which users will become high-frequency customers, used for both retention campaigns and acquisition spend allocation.

If you want to understand the foundational concepts behind most of these surfaces, the introduction to artificial intelligence and machine learning covers the building blocks before you dive into company-specific implementations.

Three Hiring Tracks: Roles, Pay Bands, and Selection Stages

All three tracks at Eternal are based in Gurugram (HQ) and Bangalore. The Eternal careers portal at zomato.com/careers lists current openings under the Eternal Limited brand for all positions below.

The CTC figures in the table below are aggregator estimates; Eternal does not officially disclose compensation bands.

Role	Estimated CTC	Selection sequence	Notes
Software Development Engineer (SDE)	₹20-35 LPA	Online coding assessment → 2 technical interviews (DSA + system design) → hiring manager round	Freshers eligible; all product lines
ML/AI Engineer, Central AI Team	₹30-60 LPA	Work demo submission → ML system design round → coding round (Python, ML algorithms) → leadership round	Freshers with strong LLM or RL research backgrounds considered
Data Scientist	₹18-35 LPA	Case study + take-home assignment → technical interview (SQL, Python, ML, product analytics) → culture and values round	Spans Blinkit, Zomato, Hyperpure, and District teams

A few points worth noting about the selection sequence for the ML/AI Engineer track. The first stage is a work demo submission, not an aptitude test. That is an unusual first filter for a consumer tech company and signals what the AI team actually values. For the SDE track, the technical interviews test DSA and system design in the standard two-round format common across product companies. The Data Scientist track is the heaviest on SQL and product analytics thinking.

Demo Over Resume: What Eternal’s AI Team Explicitly Asks For

The Times of India coverage of Eternal’s AI team expansion makes one thing explicit: the team asks candidates to showcase actual work rather than conventional resumes. This is not recruiting boilerplate. The first selection stage for the ML/AI Engineer track is a work demo submission.

What “actual work” looks like in practice:

• A GitHub repo with a fine-tuned language model, even on a small dataset
• A deployed recommender system or ranking experiment, even at toy scale
• A Kaggle competition result with documented model iterations
• A technical writeup of an ML system design decision, with code

You don’t need to replicate Zomato’s production ranking system. You need to demonstrate that you understand the problem and can ship something. A small-scale restaurant recommender built on open data (using Zomato’s public Kaggle dataset, for instance) is a legitimate demo. A fine-tuning experiment with an open-source LLM, deployed as an API endpoint, qualifies.

The practical implication for a final-year student: start building something demonstrable before you apply. The coding round for the ML track tests Python and ML algorithms, not generic aptitude. If your GitHub is empty and your resume lists only course projects, the demo-first hiring filter at Eternal’s AI team is a hard wall.

For SDEs applying to product lines outside the central AI team, the selection sequence is closer to standard product-company hiring: a DSA-heavy OA followed by two technical rounds. The system design round at the SDE level tests foundational distributed systems thinking rather than ML system design.

AI Skills for Placement Season and Beyond

Eternal’s ML problem space (recommender systems, ETA models, demand forecasting, CLV modelling) sits at the intersection of classical ML and modern LLM engineering. The foundational skills required to start contributing to these surfaces are the same ones worth building regardless of which product company you target in your placement season.

Collaborative filtering, embedding models, time-series forecasting, and basic Python ML tooling (scikit-learn, PyTorch, or JAX) are the technical floor. Above that floor, Eternal’s central AI team explicitly works on LLM fine-tuning and reinforcement learning. These aren’t placements-season additions you bolt on in a week; they take a few months of deliberate project work to get into demonstrable shape.

The 2026 AI roadmap for Indian engineering students covers which of these skills are genuinely in demand, which free resources build them fastest, and how to sequence the prep without derailing your aptitude or core CS placement work. If you’re in your pre-final year, that roadmap is the right starting point.

Eternal’s AI team selects on demos, not resumes. If recommender systems and LLM fine-tuning are the skill set they’re hiring for, TinkerLLM is where you start building that first demo: ₹299 puts real LLM API calls in your hands, and the resulting project is what goes on your GitHub before you apply to Eternal’s central AI team.