Principal Architect
Hybrid
Visionet Systems
Enterprise
Service
B2B
₹ 40-60 Lacs PA
Private Equity
Information technology
Bangalore, Karnataka, India
Post Status: Active
Permanent
5 applications
Experience: 10-16 Years
Skills
LLM
Observability
GPU Infra
GenAI
RAG
LangChain
Agentic AI
Fine-tuning
Posted 11 days ago

About the job

Visionet is a global IT services and solutions company with 25+ years of experience, serving Fortune 500 clients across BFSI, retail, and manufacturing. They are now building an enterprise-grade Generative AI platform — and this role defines the entire architecture vision for it.

As the Principal AI Architect, you will define Visionet's enterprise GenAI strategy, design scalable LLM-based systems, and drive production-grade deployments across business units. You will operate at the intersection of research, engineering, product, and executive leadership.

Key Responsibilities
• Define enterprise-wide GenAI vision, reference architectures, and scalable design patterns
• Design and oversee production-grade RAG pipelines, AI agents, copilots, and model orchestration
• Lead infrastructure design for model hosting, GPU optimization, inference performance, and LLM observability
• Implement governance frameworks covering data privacy, model security, bias mitigation, and guardrails
• Partner with product, engineering, and C-suite to translate business needs into AI solutions
• Mentor and guide senior technical teams across the organisation

Tech Stack
LLMs (Llama, Mistral) · RAG Pipelines · AI Agents · LangChain / LlamaIndex · Fine-tuning (LoRA, PEFT, RLHF) · Multimodal Architectures · GPU Infra · Inference Optimization · LLM Observability

What We're Looking For
• 10 - 15+ years in AI/ML architecture with enterprise-scale scope
• Hands-on fine-tuning experience — LoRA, PEFT, RLHF — with real production examples
• Deep experience with open-source LLMs (Llama, Mistral, and others)
• Proven experience building AI agents and autonomous workflows
• Knowledge of multimodal architectures
• Experience setting GenAI reference architectures that other teams have adopted