New opportunity

Senior Data Scientist

Grab · Bangalore, India

About this role

The Senior Data Scientist will join the Financial Services team in Bangalore to build, deploy, and manage critical data science models. This role focuses on leveraging advanced Data Science, Generative AI, and engineering to drive product-market fit across Grab's ecosystem of consumers, drivers, and merchants. Responsibilities include developing predictive models for cross-sell and personalization, utilizing LLMs for customer experiences, and managing the end-to-end lifecycle of ML models and AI agents. You will collaborate with engineering, compliance, and operations teams to deliver scalable AI solutions. Key responsibilities include: Predictive Modelling: Building models for classification, ranking, and sequential learning. MLOps & Automation: Developing pipelines for model training, deployment, and monitoring. Generative AI Development: Designing agentic solutions using prompt engineering, orchestration, and LLM fine-tuning. Ownership: Leading AI capabilities from concept to production. Required qualifications include 5+ years of experience, expert proficiency in Python, Spark, and SQL, and hands-on experience with agentic frameworks like LangGraph and LangChain. Expertise in MLOps tools such as MLflow or Kubeflow is required. Experience with real-time streaming using Flink is considered a plus. The role emphasizes engineering excellence, clean coding, and the ability to balance model performance with business constraints like latency and cost.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

PythonSparkSQLPrestoHiveMachine LearningBaggingBoostingOnline LearningRecommendation EnginesLangGraphLangSmithLangChainLLM orchestrationMLflowKubeflowTFXSageMakerFlink SQLFlink SDKGitGitHubPrompt EngineeringLoRAPEFT

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

End-to-End Model Lifecycle Ownership
The role requires managing models from concept to production, including development, deployment, and maintenance. You must demonstrate the ability to lead AI capabilities independently while ensuring engineering excellence through clean, modular code and robust version control.
Agentic Generative AI Development
A core priority is designing and deploying agentic solutions using frameworks like LangGraph and LangChain. This involves advanced tasks such as prompt engineering, LLM fine-tuning (LoRA/PEFT), and establishing guardrails for model evaluation and monitoring.
Scalable MLOps and Pipeline Automation
You must build and maintain automated MLOps pipelines using tools like MLflow or Kubeflow. The focus is on tracking production performance and creating feedback loops to ensure models remain performant and scalable within the Grab ecosystem.
Predictive Modeling for Financial Services
You will build classification, ranking, and sequential learning models to drive cross-sell and personalization strategies. This requires expert proficiency in Python, Spark, and SQL to process large-scale transaction and customer profile data.

Likely challenges to think through

Balancing Model Performance with Business Constraints
A significant challenge is the trade-off between high-performance AI models and practical constraints like latency, infrastructure cost, and scalability. Prepare by discussing how you have optimized model architectures or inference strategies to meet strict production requirements without sacrificing accuracy.
Navigating Complex Financial Data Nuances
The role involves working with financial data across multiple countries, each with unique regulatory and behavioral nuances. Prepare by highlighting your experience in feature engineering for diverse datasets and your ability to collaborate with compliance and operations teams to ensure solutions are both effective and compliant.
Real-time Integration and Streaming
While optional, the ability to implement real-time ML solutions using Flink is highly valued. The challenge lies in transitioning from batch processing to real-time streaming architectures. Prepare by reviewing your experience with Flink or similar streaming technologies and how you handle data consistency in real-time environments.

Practice your answers

Technical practice

Predictive Modeling for Financial Services

What to coverExplain your approach to building ranking and classification models for cross-selling financial products. Discuss how you handle imbalanced datasets typical in banking and how you ensure feature engineering captures meaningful transaction patterns.

Agentic Frameworks and Orchestration

What to coverDescribe your experience using LangGraph or LangChain to build agentic solutions. Focus on how you manage state, handle tool selection, and ensure reliable output when orchestrating LLM workflows.

MLOps Lifecycle Management

What to coverDetail your process for managing the end-to-end ML lifecycle using tools like MLflow or Kubeflow. Emphasize how you automate model training, deployment, and monitoring to ensure production stability.

LLM Fine-Tuning and Optimization

What to coverDiscuss your experience with PEFT and LoRA for fine-tuning LLMs. Explain how you balance model performance with inference costs and latency requirements in a production environment.

Real-time Data Processing with Flink

What to coverExplain how you would implement real-time machine learning inference using Flink SQL or SDK. Discuss the trade-offs between batch and streaming architectures for personalization use cases.

Engineering Excellence and Clean Code

What to coverHow do you ensure your machine learning code is modular, maintainable, and version-controlled? Provide examples of how you structure projects to facilitate collaboration and scalability.

Evaluation and Guardrails for Generative AI

What to coverHow do you design evaluation datasets and guardrails for LLM-based agents? Explain your methodology for monitoring model drift and ensuring the safety of AI-generated customer interactions.

Balancing Business Constraints

What to coverDescribe a scenario where you had to trade off model accuracy for latency or cost. How did you justify your design decisions to stakeholders while maintaining product-market fit?

Behavioural practice

Cross-Functional Collaboration

What to coverDescribe a time you worked with engineering, compliance, or operations teams to deploy a complex AI solution. How did you align technical goals with business requirements and regulatory constraints?

Ownership of AI Initiatives

What to coverProvide an example of leading an AI project from concept to production. How did you manage the project lifecycle, overcome technical hurdles, and ensure the final output met user needs?

Adapting to New Research

What to coverGive an example of how you have incorporated recent AI or LLM research into a practical application. How do you stay current and decide which new frameworks are worth adopting?

Problem Solving in a Flat Structure

What to coverAs an individual contributor, describe a situation where you had to solve a hard technical problem independently. How did you define your goals and ensure your work contributed to the broader team objectives?

Sharpen your knowledge

Topics to prepare

Agentic Workflow Orchestration
The role requires designing agentic solutions using frameworks like LangGraph and LangChain. Revision exercise: Design a multi-agent system architecture for a financial cross-sell scenario, defining clear roles for each agent (e.g., data retrieval, decision-making, and guardrail validation) and mapping their interaction flow.
MLOps Lifecycle Management
The position emphasizes end-to-end ownership from development to production monitoring. Revision exercise: Map out a production pipeline using MLflow or Kubeflow that includes automated model retraining, performance drift detection, and a rollback strategy for when latency or accuracy metrics deviate from business thresholds.
Predictive Modeling for Financial Services
Focus on classification, ranking, and sequential learning for banking and lending products. Revision exercise: Review techniques for handling imbalanced financial datasets and designing ranking algorithms that optimize for both user conversion and business constraints like cost and latency.

Skills to revise

LLM Fine-Tuning and Optimization
The role requires hands-on experience with LoRA and PEFT for LLM customization. Revision exercise: Practice implementing a PEFT-based fine-tuning script on a small dataset to understand how to reduce memory footprint while maintaining model performance for a specific domain task.
Real-Time Data Processing with Flink
Experience with Flink SQL and SDK is a preferred qualification for real-time ML. Revision exercise: Develop a simple Flink job that consumes a stream of transaction events and performs windowed feature aggregation to support a real-time recommendation model.
Engineering Excellence and Clean Code
The role demands modular design and version control proficiency. Revision exercise: Refactor a legacy monolithic ML script into a modular, testable codebase using Git, ensuring that data processing, model inference, and evaluation logic are decoupled and documented.

Your preparation checklist

Prepare a technical case study on agentic workflows
Develop a project walkthrough demonstrating your experience with LangGraph or LangChain. Focus on a specific use case where you orchestrated an LLM agent, implemented prompt engineering, and established evaluation guardrails for production.
Demonstrate end-to-end MLOps proficiency
Prepare to discuss a past project where you managed the full lifecycle of a model. Highlight your use of tools like MLflow or Kubeflow to automate training, deployment, and monitoring, specifically addressing how you handled model performance versus latency trade-offs.
Review predictive modeling for financial services
Refresh your knowledge on classification, ranking, and sequential learning models. Be prepared to explain how you would engineer features from transaction data to improve cross-sell or personalization outcomes in a high-scale environment.
Formulate questions on Grab's AI infrastructure
Prepare thoughtful questions regarding how the Financial Services team balances real-time streaming requirements (e.g., Flink) with batch processing for their specific banking and payments use cases.
Refine your engineering excellence narrative
Prepare examples of how you maintain clean, modular code and utilize Git/GitHub in a collaborative team setting. Be ready to discuss how you ensure scalability and cost-efficiency when deploying AI solutions into production.

Availability last checked: 18 Oct 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in India. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…