New opportunity

Machine Learning Engineer

Quadrant Technologies · Hyderabad, India

About this role

Quadrant Technologies seeks an on-site Machine Learning Engineer in Hyderabad, Telangana, India, to build, fine-tune, and deploy large language model (LLM) and generative AI solutions. The posting specifies 7–10 years of experience and describes the position as permanent.

Responsibilities

Design and deploy LLM-powered applications, including Retrieval-Augmented Generation (RAG) pipelines for enterprise use cases.
Train, fine-tune, evaluate, and optimize foundation models and domain-specific LLMs, using parameter-efficient techniques such as LoRA and QLoRA.
Build scalable applications with LangChain, LangGraph, and Hugging Face; develop deployment pipelines and production-grade AI systems.
Assess business requirements, compare technical approaches, recommend AI/ML architectures and frameworks, and contribute to system design with cross-functional teams.

Required skills:

Hands-on experience with LLMs, generative AI, RAG, fine-tuning, and LLM training, evaluation, and optimization.
Strong Python and/or Java programming skills; experience with model deployment, inference optimization, and productionization.
Understanding of vector databases, embeddings, semantic search, system design, scalability, and distributed architectures. Strong analytical reasoning, problem-solving, and the ability to select technical solutions suited to business needs are also required.

Preferred qualifications

Experience with AWS, Azure, or GCP; knowledge of MLOps, CI/CD pipelines, and containerization.
Familiarity with AI observability, monitoring, and model performance optimization, plus experience building enterprise-grade generative AI applications.

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

Skills for this role

PythonJavaMachine LearningLarge Language ModelsGenerative AIRAGLoRAQLoRALLM fine-tuningLangChainLangGraphHugging FaceFoundation modelsModel trainingModel evaluationModel deploymentInference optimizationVector databasesEmbeddingsSemantic searchSystem designScalabilityDistributed architecturesAWSAzureGCPMLOpsCI/CDContainerizationAI observability

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

Build and optimize enterprise RAG
The role explicitly calls for RAG pipelines and knowledge of vector databases, embeddings, and semantic search. Prepare to explain retrieval design, how you would evaluate answer quality, and where you would investigate weak results.
Fine-tune and evaluate LLMs
The posting names foundation and domain-specific models, LoRA, QLoRA, training, evaluation, and optimization. Prepare to explain when parameter-efficient fine-tuning is warranted and how you would compare it with other approaches.
Deliver production-grade AI systems
Responsibilities include deployment pipelines, scalable applications, and production-grade systems; inference optimization is required. Prepare a system-design account covering deployment, scalability, and operational considerations. Cloud, MLOps, CI/CD, and observability are preferred rather than required.
Choose architectures for business needs
The engineer is expected to analyze requirements, compare approaches, recommend architectures, and contribute to cross-functional system design. Prepare to show how business constraints would shape a technical recommendation.

Likely challenges to think through

Choosing RAG versus fine-tuning
A likely design trade-off, not a stated employer problem, is whether retrieval, fine-tuning, or both best serve a use case. Prepare a decision framework based on the required knowledge, evaluation results, and implementation cost.
Balancing model quality with production performance
The duties span model evaluation, inference optimization, and scalable deployment. A likely challenge is improving output quality without making an application impractical to serve. Prepare to discuss measurable quality and serving-performance trade-offs.
Turning varied requirements into a shared design
Because the role recommends approaches for diverse business use cases and collaborates across teams, a likely challenge is aligning a technical architecture with differing constraints. Prepare to explain how you would surface assumptions, compare options, and communicate a recommendation.

Practice your answers

Technical practice

Suggested practice question: How would you design an enterprise RAG application?

What to coverSuggested answer focus: Explain document ingestion, retrieval, prompt assembly, generation, and how you would evaluate the resulting answers. Discuss trade-offs between answer quality, latency, and operational complexity.

Suggested practice question: How would you choose and tune embeddings, semantic search, and a vector database?

What to coverSuggested answer focus: Describe how you would compare retrieval approaches using representative queries and relevant results. Explain how indexing and retrieval choices affect accuracy, scale, and response time.

Suggested practice question: How would you diagnose poor answers from a RAG pipeline?

What to coverSuggested answer focus: Separate retrieval failures from problems in how the model uses retrieved material. Describe targeted tests and changes you would make before deciding whether fine-tuning is warranted.

Suggested practice question: When would you use LoRA or QLoRA to fine-tune an LLM?

What to coverSuggested answer focus: Compare their resource and quality trade-offs with other model-adaptation approaches. Explain what training data and evaluation evidence you would need for a domain-specific use case.

Suggested practice question: How would you decide between a foundation model and a domain-specific LLM?

What to coverSuggested answer focus: Explain how business requirements, available data, and evaluation results would shape model selection or further training. Address the cost and maintenance implications of each approach.

Suggested practice question: How would you evaluate and optimize an LLM-powered application?

What to coverSuggested answer focus: Propose tests that reflect the intended business use case, then distinguish model-quality issues from application-level issues. Explain how you would weigh quality gains against inference cost and latency.

Suggested practice question: Where would LangChain, LangGraph, and Hugging Face fit in an LLM application?

What to coverSuggested answer focus: Describe which parts of a proposed application you would build with each framework and where simpler code might suffice. Discuss maintainability and scalability rather than assuming every framework is needed.

Suggested practice question: How would you take an LLM from development to production deployment?

What to coverSuggested answer focus: Walk through deployment-pipeline and application-code considerations using Python or Java, as appropriate. Explain how you would test changes and improve inference performance without losing sight of model quality.

Suggested practice question: How would you design an LLM service to scale?

What to coverSuggested answer focus: Discuss the boundaries between retrieval, application logic, and model inference in a distributed architecture. Explain how you would identify bottlenecks and make capacity-versus-latency trade-offs.

Suggested practice question: How would you operate and monitor a deployed LLM application?

What to coverSuggested answer focus: The posting lists cloud platforms, MLOps, CI/CD, containerization, and observability as preferred areas; describe how you would use relevant ones to release and monitor a system. Suggest signals for detecting service or model-performance degradation and how you would respond.

Behavioural practice

Suggested practice question: Tell me about a time you turned business requirements into an AI/ML approach.

What to coverSuggested answer focus: Use a real situation, your analysis of alternatives, and the recommendation you made. Explain the outcome and how business constraints influenced the decision.

Suggested practice question: Describe a time you worked across teams on a scalable system design.

What to coverSuggested answer focus: Set out the teams' differing concerns, your contribution to the design discussion, and the actions taken together. State the result and any trade-off the group accepted.

Suggested practice question: Tell me about a difficult problem in an AI application or deployment that you resolved.

What to coverSuggested answer focus: Describe the symptoms, how you narrowed down the cause, and what you changed. Give the observed result, distinguishing evidence from assumptions.

Suggested practice question: Describe a time evaluation results caused you to change a technical approach.

What to coverSuggested answer focus: Explain the original approach, what the evaluation revealed, and how you decided on a change. State the outcome, including any remaining limitation or trade-off.

Sharpen your knowledge

Topics to prepare

RAG architecture and retrieval quality
Suggested exercise: diagram an enterprise RAG flow from documents through embeddings, vector search, retrieved context, and generation; explain how you would diagnose irrelevant retrieval. Relevance: the posting calls for building and optimizing RAG pipelines and understanding embeddings, vector databases, and semantic search.
Fine-tuning strategy
Suggested exercise: compare LoRA, QLoRA, and an untuned foundation model for a domain-specific use case, stating what evidence would justify each choice. Relevance: LoRA and QLoRA are explicit requirements, alongside foundation-model and domain-specific LLM fine-tuning.
LLM evaluation and optimization
Suggested exercise: outline an evaluation plan that checks answer quality and inference performance, then explain how results would guide changes. Relevance: training, evaluation, optimization, and inference optimization are stated responsibilities.
Scalable production architecture
Suggested exercise: sketch a deployment architecture and discuss how it would handle increased requests and component failures. Relevance: the role involves production-grade AI systems, system design, scalability, and distributed architectures.
Architecture decisions from business requirements
Suggested exercise: take a hypothetical enterprise use case, compare a RAG approach with fine-tuning, and defend a recommendation with stated assumptions and trade-offs. Relevance: the engineer is expected to evaluate approaches and recommend architectures with cross-functional teams.

Skills to revise

Build a RAG application
Suggested exercise: implement a small RAG flow in Python or Java using LangChain or LangGraph, embeddings, and a vector database; inspect retrieved passages alongside generated answers. Relevance: RAG implementation and these frameworks are explicit in the posting.
Apply parameter-efficient fine-tuning
Suggested exercise: use the Hugging Face ecosystem to run a small LoRA or QLoRA fine-tuning experiment and record the setup and evaluation results. Relevance: the posting explicitly requests hands-on LoRA, QLoRA, fine-tuning, and Hugging Face proficiency.
Evaluate and optimize model behavior
Suggested exercise: write a Python or Java evaluation script for a small set of representative prompts, compare outputs across two model configurations, and investigate a quality or inference-performance regression. Relevance: programming, LLM evaluation, and optimization are required.
Develop a deployment pipeline
Suggested exercise: package an LLM application for deployment and document its startup, inference path, and a performance bottleneck you would measure. Relevance: model deployment pipelines, productionization, and inference optimization are explicit responsibilities.
Practice production operations
Optional suggested exercise: outline how you would containerize the application, run a CI/CD check, and monitor model performance on one of AWS, Azure, or GCP. Relevance: containerization, CI/CD, cloud platforms, and AI observability are listed as preferred qualifications, not requirements.

Your preparation checklist

Build a role-evidence map
The posting specifies 7–10 years of experience and hands-on LLM work. Prepare a one-page map of relevant projects, your responsibilities, technical decisions, and measurable outcomes; leave gaps unclaimed.
Prepare a RAG project walkthrough
The role calls for enterprise RAG pipelines. Outline a project or design exercise covering ingestion, embeddings, vector search, retrieval quality, evaluation, and trade-offs; identify where LangChain or LangGraph fits, if applicable.
Explain a fine-tuning decision
The posting names foundation and domain-specific models, LoRA, and QLoRA. Prepare a comparison of fine-tuning versus other approaches, including data, compute, evaluation criteria, and optimization choices.
Ready a code-level example
Python and/or Java plus Hugging Face are listed. Select a relevant implementation you can explain: its structure, tests, failure handling, and the parts you personally built. Use a small illustrative example if no shareable code is available.
Practice a business-to-architecture exercise
The role involves assessing requirements and recommending scalable designs with cross-functional teams. Sketch an LLM application architecture, compare two viable approaches, and state the assumptions and questions driving your choice.
Trace a model into production
The posting calls for deployment pipelines, inference optimization, and production-grade systems; cloud, CI/CD, containers, and observability are preferred. Prepare a deployment diagram and explain how you would test, monitor, and improve it, distinguishing experience from proposed design.
Bring targeted employer questions
Ask which enterprise use cases matter most, how LLM and RAG quality is evaluated, what deployment constraints apply, and how engineering teams share architecture decisions. These are questions to clarify, not claims about Quadrant Technologies' processes.

Availability last checked: 5 Nov 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…