New opportunity

Senior Solutions Architect, Generative AI

NVIDIA · Bengaluru, India

About this role

NVIDIA is seeking a dynamic and experienced Generative AI Solution Architect to join the AI Solutions team. This role focuses on architecting and delivering cutting-edge solutions leveraging NVIDIA's generative AI technologies, with a specific emphasis on Speech AI, LLMs, and Agentic AI.

Responsibilities

Architect end-to-end generative AI solutions focusing on Speech AI, LLMs, Agentic, and RAG workflows.
Collaborate with customers to address language-related business challenges and design tailored solutions.
Support pre-sales activities, including technical presentations and demonstrations of AI capabilities.
Lead workshops and design sessions to define generative AI solutions.
Lead the training and optimization of LLMs using NVIDIA hardware and software.
Design and implement RAG-based workflows for enhanced content generation and retrieval.
Provide technical leadership and guidance on best practices for LLM training and RAG implementation.

Required Qualifications

B.Tech, Master's, or Ph.D. in Computer Science, Artificial Intelligence, or equivalent.
8+ years of hands-on experience in a technical role focusing on generative AI, specifically training LLMs and Speech AI.
Proven track record of deploying and optimizing LLM and Speech AI models for production inference.
In-depth understanding of state-of-the-art language models (e.g., GPT-3, BERT).
Expertise in training/fine-tuning LLMs using TensorFlow, PyTorch, or Hugging Face.
Proficiency in model deployment and optimization for GPU platforms.
Strong knowledge of GPU cluster architecture and parallel processing.
Excellent communication and presentation skills.

Preferred Qualifications

Ability to optimize models for inference speed, memory efficiency, and resource utilization.
Familiarity with containerization (Docker) and orchestration (Kubernetes).
Deep understanding of distributed computing concepts.
Hands-on experience with NVIDIA GPU technologies and cluster management.

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

Skills for this role

Generative AISpeech AILarge Language Models (LLMs)Agentic AIRAGTensorFlowPyTorchHugging Face TransformersGPU cluster architectureParallel processingModel deploymentModel optimizationDockerKubernetesDistributed computing

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 Generative AI Architecture
The role requires designing and delivering comprehensive solutions covering Speech AI, LLMs, and Agentic/RAG workflows. You must demonstrate the ability to translate complex customer business challenges into technical architectures that leverage NVIDIA's specific hardware and software stack.
LLM Training and Optimization Leadership
A core priority is leading the training, fine-tuning, and production-level optimization of LLMs. This involves deep expertise in frameworks like PyTorch or TensorFlow and the ability to maximize model performance, inference speed, and memory efficiency on GPU clusters.
Technical Pre-Sales and Stakeholder Engagement
You will act as a technical bridge between NVIDIA and its customers. This requires leading workshops, conducting technical demonstrations, and providing guidance on best practices to both technical and non-technical stakeholders to drive adoption of NVIDIA generative AI technologies.

Likely challenges to think through

Balancing Customization with Scalability
Customers often require highly tailored AI solutions, but you must ensure these designs remain scalable and efficient on GPU clusters. Prepare by being ready to discuss trade-offs between model accuracy, inference latency, and hardware resource utilization in distributed environments.
Translating Complex AI Concepts for Diverse Audiences
As a Solutions Architect, you will face the challenge of explaining advanced concepts like RAG workflows or parallel processing to stakeholders with varying levels of technical expertise. Prepare by practicing how to simplify high-level architectural decisions without losing the technical rigor required for implementation.
Rapid Evolution of Generative AI Technologies
The field of LLMs and Speech AI moves quickly, and you are expected to stay at the forefront. Prepare by being ready to discuss how you keep your knowledge current and how you apply the latest research or architectural patterns to solve real-world production problems.

Practice your answers

Technical practice

Optimizing LLM Inference on GPU Clusters

What to coverDiscuss strategies for maximizing throughput and minimizing latency when deploying LLMs on NVIDIA GPU clusters. Address techniques such as model quantization, tensor parallelism, and pipeline parallelism, and how these interact with memory bandwidth and compute utilization.

Designing Scalable RAG Workflows

What to coverExplain the architectural components required for a production-grade Retrieval-Augmented Generation (RAG) system. Focus on the trade-offs between different vector database choices, embedding models, and the impact of retrieval strategies on overall generation quality.

Training and Fine-tuning Strategies

What to coverDescribe your approach to fine-tuning large language models using frameworks like PyTorch or Hugging Face. Discuss how you manage distributed training across multiple GPUs and the specific challenges associated with memory efficiency during the training process.

Speech AI Integration Challenges

What to coverDetail the complexities involved in integrating Speech AI models into end-to-end generative AI pipelines. Address how you handle real-time processing requirements and the synchronization between speech recognition and downstream LLM reasoning.

GPU Architecture and Parallel Processing

What to coverExplain how you leverage GPU cluster architecture to accelerate AI workloads. Discuss the importance of understanding interconnects (like NVLink) and how parallel processing concepts are applied to both training and inference tasks.

Agentic AI Workflow Implementation

What to coverDefine the architectural requirements for building agentic AI systems that can perform multi-step reasoning and tool usage. Discuss how to manage state, error handling, and the orchestration of multiple model calls within a cohesive workflow.

Model Deployment and Containerization

What to coverDiscuss the role of Docker and Kubernetes in deploying generative AI models at scale. Explain how you ensure reproducibility, manage dependencies, and handle resource scheduling for GPU-intensive workloads in a production environment.

Performance Benchmarking and Evaluation

What to coverDescribe your methodology for evaluating the performance of a deployed AI solution. Focus on metrics for inference speed, memory footprint, and the qualitative assessment of model outputs in a RAG-based context.

Behavioural practice

Translating Business Needs to Technical Solutions

What to coverDescribe a situation where you had to translate a customer's vague business challenge into a concrete generative AI architecture. Explain how you balanced technical feasibility with the customer's specific requirements.

Leading Technical Workshops

What to coverProvide an example of a time you led a technical workshop or design session for a diverse audience. Explain how you adapted your communication style to ensure both technical and non-technical stakeholders understood the proposed AI solution.

Collaborating with Engineering Teams

What to coverShare an experience where you provided feedback to an engineering team based on customer interactions. Detail how this feedback influenced the evolution of a product or technology and the outcome of that collaboration.

Handling Technical Objections in Pre-sales

What to coverDescribe a scenario where you faced skepticism regarding the viability or performance of an AI solution during a pre-sales demonstration. Explain the steps you took to address these concerns and the final result.

Sharpen your knowledge

Topics to prepare

RAG and Agentic AI Workflows
Review the architectural patterns for Retrieval-Augmented Generation (RAG) and autonomous agentic systems. Practice designing end-to-end workflows that integrate external knowledge retrieval with LLM reasoning. This is critical as the role specifically requires leading design sessions for these complex systems.
GPU Cluster Architecture and Distributed Computing
Study the principles of parallel processing, distributed training, and GPU cluster management. Focus on how to scale model training and inference across multiple nodes. This is essential for providing technical leadership on NVIDIA's hardware platforms.

Skills to revise

LLM Training and Optimization
Practice fine-tuning models using PyTorch or Hugging Face, with a focus on optimizing for inference speed, memory efficiency, and resource utilization. This directly addresses the requirement to lead training sessions and optimize models for production.
Containerization and Orchestration
Develop proficiency in deploying AI models using Docker and Kubernetes. Create a sample deployment pipeline that demonstrates scalable inference on GPU-enabled clusters. This is a preferred qualification for ensuring efficient model delivery.
Technical Communication and Pre-sales
Prepare to explain complex AI architectures to non-technical stakeholders. Practice conducting a mock workshop or technical demonstration of a RAG-based solution. This is vital for the pre-sales and customer-facing aspects of the role.

Your preparation checklist

Prepare a technical case study on LLM/Speech AI deployment
Develop a concise walkthrough of a production-grade generative AI project you led. Focus on the specific challenges of training or fine-tuning models using PyTorch or TensorFlow, and explain how you optimized these models for inference speed and memory efficiency on GPU hardware.
Document your experience with GPU cluster architecture
Create a summary of your experience managing or architecting GPU clusters. Highlight your ability to leverage parallel processing and distributed computing concepts to scale training and inference workflows, as this is critical for NVIDIA's infrastructure-focused role.
Design a RAG-based workflow architecture
Draft a high-level design for an end-to-end RAG (Retrieval-Augmented Generation) system. Be prepared to explain your choice of vector databases, retrieval strategies, and how you ensure the system is scalable and efficient within a containerized (Docker/Kubernetes) environment.
Refine your technical communication for pre-sales
Prepare to explain complex generative AI concepts, such as agentic workflows or model optimization, to both technical engineering teams and non-technical business stakeholders. Practice articulating how these solutions directly address specific customer business challenges.
Formulate strategic questions for the interviewers
Prepare thoughtful questions regarding the current challenges NVIDIA customers face when scaling generative AI, the evolution of their internal AI software stack, or how the solutions team collaborates with engineering to influence future product roadmaps.

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…