# AI Architect

> Unison Group · Chennai, India (Hybrid) · Contract · Posted 2026-05-05

**Workplace:** hybrid

**Department:** Manish

## Description

**Responsibilities**

**Agentic AI Systems**

-   Design reusable patterns for Agentic AI systems including RAG, Multi-Agent Orchestration, and Human-in-the-loop systems
-   Define how different agents communicate, share state, and hand off tasks to one another
-   Architect long-term and episodic memory layers using Vector Databases, embedding pipelines, and knowledge graphs
-   Decide when to use high-reasoning models vs. worker models to optimise cost and performance
-   Predict and control token usage; architect systems with semantic caching to prevent redundant LLM spend
-   Set architectural standards for explainability, auditability, and guardrails to prevent hallucinations and bias
-   Ensure data governance, privacy compliance, and responsible AI practices across all systems

**AI Infrastructure & MLOps**

-   Design scalable AI infrastructure including model serving, inference architecture, AI microservices, and APIs
-   Architect distributed systems supporting AI workloads
-   Define MLOps and CI/CD pipelines for AI systems
-   Architect containerised and cloud-native deployments; design monitoring and observability for AI services
-   Optimise for cost, performance, and scalability across the AI stack

**Enterprise AI & Agentic Architecture**

-   Architect enterprise-scale Agentic AI frameworks using LangGraph, Model Context Protocol (MCP), multi-agent orchestration frameworks, and memory-driven AI systems
-   Design and implement RAG pipelines (Hybrid RAG, Graph-RAG), embeddings pipelines (open-source and enterprise models), prompt orchestration, guardrails, and fine-tuning pipelines (PEFT, LoRA, domain adaptation)
-   Build secure LLM deployments across on-prem, air-gapped, and cloud-agnostic environments
-   Define LLMOps lifecycle covering evaluation harness, hallucination detection, observability (tracing, telemetry), and model governance
-   Hands-on experience with agentic AI frameworks — LangChain, LlamaIndex, AutoGen, CrewAI

**Data Platform & Lakehouse Engineering**

-   Design and govern modern data platforms built on Medallion (Bronze-Silver-Gold) architecture with Delta tables and ACID transactional layers
-   Architect multi-tenant platforms with cost governance and data mesh or federated data architecture patterns
-   Work across the core stack: Databricks, Apache Spark (batch & streaming), Delta Live Tables, Apache Druid, Dremio, Kubeflow Pipelines, Airflow
-   Drive schema evolution and versioning, metadata and lineage management, data quality frameworks, dimensional modelling for analytics, and Kafka-based streaming ingestion

**Advanced AI/ML & Deep Learning**

-   Architect ML systems using TensorFlow, PyTorch, Scikit-Learn, XGBoost, LSTM, CNN, Transformer models, and Vision-Language Models (VLMs)
-   Design time-series forecasting and anomaly detection solutions for industrial telemetry

**Cloud, Infrastructure & DevOps**

-   Cloud-native AI architecture on Azure and AWS
-   Containerisation using Docker and Kubernetes (Helm, Operators)
-   Infrastructure as Code using Terraform
-   CI/CD for ML pipelines with secure DevSecOps integration
-   Hybrid and on-prem deployments under compliance constraints

**Databases, Graph & Vector Systems**

-   RDBMS: PostgreSQL; NoSQL: MongoDB
-   Graph Databases: Neo4j for ontology and knowledge graph modelling
-   Vector Databases: Pinecone, FAISS, Milvus, and enterprise vector DB solutions
-   Context modelling and semantic search frameworks

## Requirements

**Required Experience**

-   15+ years in Data, AI, and Platform Engineering
-   5+ years in an AI Architecture leadership role
-   Proven delivery of enterprise-scale AI platforms in production environments
-   Experience in industrial or engineering AI ecosystems
-   Strong background in distributed systems and scalable data processing
-   B.Tech/BE in Computer Science or related field; M.Tech/MS in Data Science or AI preferred

## Apply

[Apply at Unison Group](https://apply.workable.com/unisongroup/j/8D4AB281C6/apply)

---
Powered by [Workable](https://www.workable.com)