AI Architect

Larsen & Toubro; Chennai District; On-site; INR 4,000,000 - 7,000,000; Full time

Job summary

Larsen & Toubro in India seeks a seasoned AI Cloud Platform Architect to design end-to-end AI cloud ecosystems with focus on security, cost-efficiency, and performance.

The role requires translating complex requirements into scalable, production-grade AI platforms, GPU clusters, and Kubernetes-based orchestration, with a track record in large-scale deployments.

Qualifications

• BE/BTech or equivalent in CS/EC.
• 15–20 years IT experience with minimum 5 years in AI platform.
• Proven expertise in production-grade AI/ML architectures.
• Strong client engagement and cross-functional collaboration.

Responsibilities

• Translate business requirements into scalable AI/GenAI architectures with GPU clusters.
• Design end-to-end AI Cloud platforms for DL workloads and distributed training.
• Architect HPC topologies using InfiniBand RoCE v2 for low latency.
• Right-size GPUs, CPUs, memory and NVMe for client proposals.
• Establish IaC and CI/CD pipelines for reproducible deployments.

Skills

NVIDIA GPU Architecture

Kubernetes

Slurm

OpenShift

Python

PyTorch

TensorFlow

LangChain

LangGraph

LLM APIs

Pinecone

FAISS

Chroma

RLHF

GPU Cluster

Education

BE/BTech in CS/EC

Tools

Docker

CI/CD pipelines

Git workflows

OpenStack

Job description

Job Purpose

Designs and architect end-to-end AI Cloud platforms with a focus on security, cost-efficiency, and performance. This position involves direct client engagement to translate requirements into technical solutions, encompassing GPU infrastructure rightsizing and optimal model selection. We are looking for a cloud expert with a demonstrated ability to transition complex AI models from concept to large-scale production. The ideal candidate brings extensive experience in AI/Cloud ecosystems and a successful track record of architecting and managing production-grade, large‑scale AI platforms.

Key Responsibilities

• Translate business requirements into scalable, high-performance AI/GenAI architectures featuring NVIDIA GPU clusters
• Design end-to-end AI Cloud and next-generation platforms optimized for deep learning workloads and distributed training.
• Architect HPC cluster topologies utilizing high-speed InfiniBand (NDR/HDR) and RoCE v2 interconnects for low‑latency communication.
• Right‑size platform components, including GPUs, CPUs, memory and NVMe storage for comprehensive client proposals.
• Architect distributed training and inference environments optimized for MPI frameworks and workload scheduling via Slurm.
• Design scalable container orchestration platforms using Kubernetes and Kubeflow to manage AI workloads.
• Propose optimized inference strategies using vLLM, Triton, and TensorRT‑LLM to meet specific latency and throughput KPIs.
• Have experience on RAG systems and multi‑agent orchestration frameworks like LangGraph and agentic ecosystems.
• Develop private AI cloud environments focused on data sovereignty and regulatory compliance, such as the India DPDP Act.
• Define integration strategies for LLMs and open‑source models within existing enterprise data systems, APIs, and knowledge graphs.
• Establish reference architectures for CI/CD/CT pipelines and automated model retraining workflows to ensure reproducibility.
• Implement automation and observability frameworks for monitoring GPU utilization, performance tuning, and failure handling.
• Drive technical validation through proof‑of‑concept engagements, focusing on scalability and performance benchmarks for LLM training.
• Establish Infrastructure‑as‑Code (IaC) practices to ensure reproducible and reliable cluster deployments.
• Collaborate with C‑suite stakeholders and cross‑functional teams to drive technical decision‑making, innovation, and roadmap alignment.

Qualifications and Experience

Educational Qualifications: BE/B‑Tech or equivalent with Computer Science or Electronics & Communication

Relevant Experience: 15–20 years of IT experience with a minimum of 5 years in AI platform.

Core AI/ML Expertise

• Strong experience in Nvidia, Intel, Google GPU Architecture, InfiniBand
• Strong expertise in Kubernetes, Slurm and OpenShift
• Good experience in Python, PyTorch and TensorFlow
• Good knowledge on LangChain, LangGraph
• Deep understanding of Transformers, Attention mechanisms, Diffusion, MoE
• Knowledge of RLHF, Pinecone, FAISS, Chroma, OpenAI, VLLM
• Expertise in RAG and agentic AI workflows
• Knowledge of high‑performance storage (Lustre, PFS, Object NVMe)
• Good Knowledge with NVIDIA architectures (Hopper, Blackwell)

Soft Skills

• Strong problem‑solving and analytical thinking
• Excellent communication and stakeholder management
• Ability to influence leadership and drive strategic decisions
• Innovation mindset with focus on enterprise impact

Preferred Experience

• Currently in AI / Cloud Presales team
• Should be able to right‑size infra and choose right GPU model as per client requirement
• Hands‑on with Python, vector DBs (Pinecone, FAISS, Chroma), and LLM APIs (OpenAI, Anthropic).
• Solid understanding of cloud‑native architecture (OpenStack, KVM, Azure/AWS/GCP), microservices, Kubernetes, serverless, API gateways.
• Good knowledge of deep learning experience: CNNs, RNNs/LSTMs, Transformers, and attention mechanisms.
• Proficiency in Python for ML: NumPy, pandas, scikit‑learn, and frameworks such as PyTorch or TensorFlow.
• Experience in integrating LLMs (GPT, Claude, Gemini, LLaMA, Mistral) into applications.
• Prompt engineering skills: zero‑shot, few‑shot, chain‑of‑thought, ReAct, and structured output patterns.
• Experience building RAG systems: document chunking, embedding models, vector search, and retrieval optimization.
• Understanding of AI agent patterns, tool use, and agentic workflows.
• Familiarity with Docker, CI/CD pipelines, and Git‑based workflows.
• Strong communication, stakeholder management, and solution design skills.