New opportunity

Sr. Data Scientist, AI Engineering

Trinity Partners India LLP · Bangalore, India

About this role

The Sr. Data Scientist, AI Engineering will architect and implement advanced AI solutions, focusing on RAG pipelines, agentic AI systems, and LLM-driven applications for enterprise use cases. Key responsibilities include designing prompt engineering frameworks, managing context and knowledge-grounding to optimize LLM performance, and collaborating with cross-functional teams to build production-grade GenAI microservices and APIs. The role involves evaluating open-source and proprietary LLMs, designing data pipelines for content ingestion and vector retrieval, and defining architecture standards for scalable GenAI platforms. You will lead PoCs, benchmark model performance, and manage the transition from prototype to production while contributing to AI strategy and cost optimization. Required Qualifications: Bachelor’s or Master’s degree in Computer Science, AI, Data Engineering, or a related field. A minimum of 5 years of total experience is required, including at least 3 years in AI/ML or NLP systems design. Candidates must have proven experience with LLM-based solutions, RAG architectures, and prompt orchestration frameworks, along with strong Python programming skills and familiarity with frameworks like LangChain, LangGraph, LlamaIndex, or Transformers. Hands-on knowledge of vector databases and knowledge graph systems is essential, as is experience deploying AI workloads on cloud platforms (GCP, Azure, AWS) and understanding MLOps/GenAIOps practices. Preferred Qualifications: Experience with multimodal LLMs, agentic reasoning, and tool-using AI agents. Exposure to pharma, life sciences, or regulated enterprise domains is highly valued, as is a history of contributing to open-source AI frameworks.

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

Skills for this role

PythonRAGLLMPrompt EngineeringLangChainLangGraphLlamaIndexTransformersMilvusPineconeWeaviateChromaNeo4jRDFPostgreSQLpgvectorGCPAzureAWSMLOpsGenAIOpsAgentic AI

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

Architecting Scalable RAG and Agentic Pipelines
The role requires designing and implementing end-to-end RAG pipelines and agentic AI systems. You must demonstrate the ability to build robust data ingestion, indexing, and vector retrieval workflows that support enterprise-grade LLM applications.
Productionizing GenAI Solutions
Moving beyond PoCs to production is a core expectation. You must be prepared to discuss the full lifecycle of GenAI microservices, including model benchmarking, cost optimization, and the implementation of MLOps/GenAIOps practices to ensure reliability in a regulated environment.
Prompt Engineering and Context Management
You are expected to define frameworks for prompt orchestration and knowledge grounding. This involves optimizing LLM performance through sophisticated context management and ensuring the system provides accurate, grounded responses for complex enterprise use cases.

Likely challenges to think through

Balancing Innovation with Regulatory Constraints
Working in the life sciences domain often involves strict data governance and compliance requirements. You may face difficulty when trying to implement cutting-edge agentic AI while adhering to the rigorous security and validation standards required in pharmaceutical or biotech industries.
Managing Model Lifecycle and Cost at Scale
As you transition from prototype to production, you will likely encounter the trade-off between model performance and operational costs. Prepare to discuss how you evaluate proprietary versus open-source models and how you implement observability to monitor performance and budget effectively.

Practice your answers

Technical practice

Designing Scalable RAG Pipelines

What to coverExplain your approach to building a production-grade RAG pipeline. Discuss how you handle unstructured data ingestion, chunking strategies, and the trade-offs between different vector retrieval methods to ensure high accuracy and low latency.

Prompt Engineering and Context Management

What to coverDescribe your methodology for designing prompt frameworks that maintain consistency across enterprise applications. Address how you manage context windows and implement knowledge-grounding to minimize hallucinations in LLM outputs.

Agentic AI and Tool Orchestration

What to coverDiscuss your experience with agentic reasoning and tool-using AI agents. Explain how you design systems where an LLM can autonomously select and execute tools to solve complex, multi-step user queries.

Vector Databases vs. Knowledge Graphs

What to coverCompare the use cases for vector databases (e.g., Milvus, Pinecone) versus knowledge graphs (e.g., Neo4j, RDF) in RAG architectures. Explain how you would integrate both to improve retrieval precision for complex domain-specific queries.

GenAIOps and Model Observability

What to coverDetail your strategy for implementing GenAIOps in a production environment. Focus on how you monitor LLM performance, track model drift, and ensure observability for deployed microservices.

Benchmarking and Model Evaluation

What to coverHow do you evaluate the performance of proprietary versus open-source LLMs for a specific enterprise use case? Discuss the metrics you use to benchmark accuracy, cost, and inference speed during the PoC phase.

Transitioning from Prototype to Production

What to coverDescribe the challenges of moving a GenAI prototype to a scalable production system. Focus on architecture standards, cost optimization, and ensuring the system meets enterprise governance requirements.

Cloud Deployment of AI Workloads

What to coverExplain your experience deploying AI workloads on cloud platforms like GCP, AWS, or Azure. Discuss how you handle containerization, API management, and scaling strategies for high-traffic GenAI applications.

Behavioural practice

Cross-functional Collaboration

What to coverDescribe a situation where you had to collaborate with non-technical stakeholders or cross-functional teams to deliver an AI solution. How did you communicate technical constraints and align the project with business goals?

Managing Technical Trade-offs

What to coverProvide an example of a time you had to make a difficult trade-off between model performance, latency, and cost. What was your decision-making process, and what was the final outcome?

Handling Ambiguity in AI Projects

What to coverTell me about a time you led a PoC for an AI project with unclear requirements. How did you define the scope, manage expectations, and ensure the project provided measurable value?

Problem-Solving in Regulated Domains

What to coverIf you have experience in pharma or regulated industries, describe a challenge you faced regarding data privacy or compliance. How did you design your AI architecture to address these specific constraints?

Sharpen your knowledge

Topics to prepare

RAG Pipeline Architecture
Focus on the end-to-end design of retrieval-augmented generation systems, specifically addressing unstructured data ingestion, indexing strategies, and vector retrieval optimization. Exercise: Draft a high-level architecture diagram for a RAG system that handles document chunking, embedding generation, and context-grounding to minimize hallucinations.
Agentic AI and Tool Orchestration
Understand the design patterns for agentic systems that utilize external tools and reasoning loops to complete complex tasks. Exercise: Review the LangGraph documentation to understand how to implement stateful, multi-agent workflows and define a plan for a PoC that uses an agent to query a knowledge graph.
GenAIOps and Model Evaluation
Study the lifecycle management of LLM applications, including benchmarking performance, cost optimization, and observability in production. Exercise: Define a set of metrics to evaluate LLM response quality (e.g., faithfulness, relevance) and outline a strategy for monitoring these metrics in a production microservice.

Skills to revise

Vector Database Implementation
Gain hands-on experience with vector databases like Milvus, Pinecone, or pgvector. Exercise: Build a small prototype that ingests a set of documents into a vector store and performs a similarity search to retrieve context for a prompt.
Prompt Orchestration Frameworks
Develop proficiency in using frameworks like LangChain or LlamaIndex to manage prompt templates and chains. Exercise: Create a Python script using LangChain to chain multiple LLM calls, where the output of one step informs the prompt of the next.
Knowledge Graph Integration
Familiarize yourself with integrating structured knowledge graphs (e.g., Neo4j) with LLM workflows for improved grounding. Exercise: Map a simple domain model into a graph structure and write a query to retrieve relevant nodes to augment an LLM prompt.

Your preparation checklist

Prepare a RAG system architecture walkthrough
Develop a technical summary of a RAG pipeline you have built, focusing on your choice of vector database (e.g., Milvus, Pinecone), indexing strategy, and how you handled context grounding or retrieval optimization to improve LLM accuracy.
Demonstrate agentic AI and orchestration expertise
Prepare to discuss your experience with frameworks like LangChain or LangGraph. Be ready to explain how you have designed agentic workflows, managed tool-use, or implemented multi-step reasoning for complex enterprise tasks.
Review MLOps and GenAIOps best practices
Outline your approach to moving GenAI prototypes to production, specifically addressing model evaluation, observability, cost optimization, and CI/CD pipelines for AI workloads on cloud platforms like AWS, GCP, or Azure.
Connect AI solutions to life sciences domains
Research Trinity's focus on life sciences commercialization. Prepare to discuss how you would adapt AI solutions for regulated environments, such as ensuring data privacy, compliance, or handling domain-specific unstructured data.
Formulate strategic questions for the interviewers
Prepare thoughtful questions regarding the company's current GenAI platform architecture, how they balance proprietary versus open-source model usage, and the specific challenges they face in scaling AI microservices for their clients.

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…