Job Title: AI Architect
Location: India – Pune, Bangalore
Experience Required: 12+ years
Engagement: [Full time]
Project Type: High-visibility R&D; initiative in financial research & data intelligence

Job Summary:
We are seeking an experienced, hands‑on AI Architect (12+ years) to lead technical delivery and client engagement for Agentic AI platforms. The role owns end‑to‑end design, productionisation and operationalisation of LLM‑based solutions (RAG, fine‑tuning, model serving) on Azure, and will act as the technical face to the client while leading a cross‑functional delivery team.

What Success Looks like:
· Deliver production Agentic AI features that meet SLA targets for latency, throughput and reliability.
· Reduce time‑to‑value for LLM integrations via repeatable patterns, MLOps pipelines and reusable components.
· Maintain model governance, explainability and security posture appropriate for sensitive data domains.

Key Responsibilities:
· Architect and build Agentic AI systems: orchestrate agents, action executors, retrieval layers, and feedback loops.




· Design and implement LLM solutions (RAG, retrieval chains, prompt engineering, fine‑tuning/LoRA/P-tuning) for production use.
· Own model deployment, serving and scaling on Azure (Azure AI, Azure ML, AKS, container registries) and hybrid setups.
· Build MLOps & ModelOps pipelines: CI/CD for models and services, automated testing, monitoring, drift detection and rollbacks.
· Lead data pipelines for retrieval: vector stores, semantic search, indexing, embeddings, data privacy & access controls.
· Implement model explainability, confidence scoring, adversarial protections and prompt security (prompt injection mitigation).
· Define and enforce model governance: versioning, reproducibility, lineage, audit trails and compliance.
· Collaborate with product, data engineering, security, DevOps and UX to ensure integrated delivery and acceptance.
· Mentor and upskill engineers; conduct technical reviews and pair prog

📌 AI Architect (India)
🏢 LatentBridge
📍 India