New opportunity

GenAI-Engineer

Fractal Analytics Ltd. · Bengaluru, India

About this role

The GenAI Engineer will design, develop, and implement advanced solutions leveraging Large Language Models (LLMs) to solve complex business challenges. The role requires taking full ownership of initiatives, delivering end-to-end solutions, and staying current with advancements in Generative AI, RAG systems, and applied research. Responsibilities include building reusable code libraries, frameworks, and accelerators; contributing to the full software development lifecycle; and collaborating with cross-functional teams to align AI solutions with business goals. The engineer will design scalable, production-grade AI systems, define architecture patterns for LLM applications, and ensure reliability through evaluation frameworks, guardrails, and responsible AI practices. Key duties involve translating ambiguous business problems into structured AI approaches, driving solution success metrics, and mentoring junior team members. Required qualifications include 5-8 years of experience as a Data Scientist with deep expertise in GenAI, NLP, and LLM technologies, including hands-on experience in ML model building, training, and fine-tuning LLMs. Candidates must possess a strong technical background in machine learning and deep learning, along with practical experience using modern AI frameworks. Proficiency in SaaS-based LLMs, agentic frameworks (e.g., CrewAI, LangGraph), and advanced RAG strategies is essential. Experience with the Databricks ecosystem and MS Copilot Studio is highly valued.

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

Skills for this role

Generative AINLPLLMMachine LearningDeep LearningLangChainLlamaIndexVector databasesPrompt engineeringAzure OpenAIMultimodal modelsClaude OpusGPT-3.5 TurboGPT-4CrewAIAutoGenLangGraphn8nLangFlowSmolAgentsSemantic KernelRAGDatabricksMS Copilot StudioMCP server

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 Solution Ownership
The role requires taking full responsibility for the entire software development lifecycle, from initial design and architecture to deployment and maintenance. You must demonstrate the ability to translate ambiguous business problems into structured, production-grade AI systems with minimal supervision.
Advanced RAG and Agentic Framework Implementation
A core priority is designing and building sophisticated LLM applications using advanced RAG strategies (e.g., hybrid, multi-hop, hierarchical) and agentic frameworks like LangGraph or CrewAI. You must be capable of defining architecture patterns that balance latency, cost, and accuracy.
Scalable and Reusable AI Engineering
You are expected to build reusable code libraries, frameworks, and accelerators that can be leveraged across multiple client use cases. This involves contributing to internal IP creation and ensuring that all AI solutions are modular, maintainable, and scalable within the Databricks ecosystem.
Responsible AI and Quality Assurance
Ensuring the reliability of LLM systems is critical. You must implement robust evaluation frameworks, guardrails, and hallucination mitigation strategies to ensure safe and responsible AI deployment, while also mentoring junior team members on these quality standards.

Likely challenges to think through

Balancing Technical Complexity with Business Value
You will face the challenge of translating vague business requirements into measurable AI success metrics. Prepare by practicing how to explain technical trade-offs—such as choosing between a SaaS LLM and a fine-tuned model—to non-technical stakeholders in terms of cost, performance, and business impact.
Managing Rapidly Evolving AI Tooling
The requirement to stay current with advancements in GenAI and agentic frameworks is significant. Prepare by being ready to discuss how you evaluate new tools (like new MCP servers or frameworks) for production readiness versus experimental use, and how you decide when an LLM is not the right tool for a specific problem.

Practice your answers

Technical practice

Advanced RAG Architectures

What to coverExplain how you would implement hybrid retrieval and multi-hop RAG to improve accuracy in complex document retrieval tasks. Discuss the trade-offs between hierarchical and contextual retrieval strategies regarding latency and retrieval precision.

Agentic Framework Implementation

What to coverCompare the use of agentic frameworks like LangGraph or CrewAI for building autonomous workflows. Describe how you manage state, tool selection, and error handling when designing multi-agent systems.

LLM Fine-Tuning and Optimization

What to coverDetail your process for fine-tuning an open-source LLM from the ground up versus using SaaS-based models. Address how you handle data preparation, compute resource management, and evaluation of model performance post-training.

Production-Grade AI Reliability

What to coverHow do you design evaluation frameworks and guardrails to mitigate hallucinations and bias in production LLM applications? Suggest methods for implementing fallback strategies when an LLM fails to provide a reliable response.

Cost-Performance Trade-offs

What to coverWhen designing scalable AI systems, how do you balance latency, cost, and accuracy? Provide an example of how you would decide between using a high-performance model like Claude Opus versus a more efficient model like GPT-3.5 Turbo for a specific business use case.

Databricks Ecosystem Integration

What to coverDescribe your experience integrating LLM pipelines within the Databricks ecosystem, specifically using tools like Databricks Genie or AgentBricks. Explain how these tools facilitate the deployment and monitoring of AI solutions.

Prompt Engineering Techniques

What to coverDiscuss the application of Chain-of-Thought (CoT) and ReAct prompting in complex agentic workflows. Explain how these techniques improve reasoning capabilities and when they might introduce unnecessary latency.

Identifying Non-LLM Alternatives

What to coverDescribe a scenario where you would recommend against using an LLM for a business problem. What classical NLP or machine learning approaches would you propose as a more efficient or reliable alternative?

Behavioural practice

Translating Ambiguous Requirements

What to coverDescribe a time you had to translate an ambiguous business problem into a structured AI solution. Explain the steps you took to define success metrics and align the technical approach with stakeholder expectations.

Mentorship and Quality Standards

What to coverProvide an example of how you have mentored junior team members on AI best practices or code quality. How do you ensure that your team maintains high standards during the full software development lifecycle?

Stakeholder Communication

What to coverTell me about a time you had to explain a complex AI architecture or trade-off to a non-technical stakeholder. How did you ensure they understood the business impact and the rationale behind your technical decisions?

Ownership of End-to-End Initiatives

What to coverDescribe a project where you took full ownership of an AI initiative from design to deployment. How did you handle unexpected challenges, and what was the final business impact of the solution?

Sharpen your knowledge

Topics to prepare

Advanced RAG Architectures
Review strategies for hybrid, knowledge graph-based, and multi-hop retrieval. Exercise: Design a RAG pipeline architecture that incorporates hierarchical retrieval and explain how you would evaluate its performance against latency and accuracy metrics.
Agentic Workflow Design
Understand the conceptual differences between autonomous agent frameworks like CrewAI and LangGraph. Exercise: Map out a multi-agent workflow for a complex business process, identifying where specific guardrails and fallback strategies are required.
Responsible AI and Governance
Study methods for hallucination mitigation, bias handling, and safety in production LLM systems. Exercise: Draft a checklist for auditing an LLM application for potential risks before deployment to a production environment.

Skills to revise

LLM Fine-Tuning and Training
Hands-on experience with training and fine-tuning models from the ground up. Exercise: Practice fine-tuning a small open-source model on a specific domain-relevant dataset and document the impact on model performance metrics.
Databricks AI Ecosystem
Proficiency in Databricks Genie, AI/BI, and AgentBricks. Exercise: Create a simple notebook in a Databricks environment that demonstrates the integration of a vector database with an LLM for a retrieval-based task.
Prompt Engineering and Function Calling
Mastering techniques like Chain-of-Thought (CoT), ReAct, and Azure OpenAI function calling. Exercise: Develop a set of prompts and function definitions that allow an LLM to interact with an external API to retrieve and process structured data.

Your preparation checklist

Prepare a RAG system architecture walkthrough
Develop a technical presentation or document detailing a production-grade RAG pipeline you have built. Focus on your approach to hybrid retrieval, multi-hop reasoning, and how you implemented evaluation frameworks to monitor hallucination and latency.
Demonstrate agentic workflow expertise
Prepare a project summary showcasing your experience with at least one agentic framework like LangGraph, CrewAI, or AutoGen. Explain how you designed the orchestration layer, managed tool usage, and handled state persistence for complex, multi-step business tasks.
Articulate your LLM fine-tuning and optimization strategy
Be ready to discuss your hands-on experience in training or fine-tuning models from the ground up. Prepare to explain your decision-making process regarding cost-performance trade-offs, such as when to use a SaaS-based model versus a fine-tuned open-source model.
Draft a framework for responsible AI and governance
Prepare a set of principles or a checklist you use to ensure reliability and safety in AI systems. This should include your methods for bias mitigation, hallucination detection, and implementing guardrails in production environments.
Formulate questions regarding business impact and scale
Prepare thoughtful questions for the interviewers about how Fractal measures the success of their AI initiatives, such as how they track productivity gains or decision accuracy, and how they approach scaling modular components across diverse client use cases.

Availability has not yet been verified. 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…