Agentic AI Architecture & Development

Cephas Consultancy Services Private Limited

India, Karnataka, Bengaluru

On-site · Full-time

Agentic AI Architect & Developer

Location: Bangalore, Karnataka, India
Positions: 2
Employment type: Full-time
Experience: 7–11 years

Design and build production-grade multi-agent systems using LangGraph as the primary orchestration framework, with knowledge of LangChain, CrewAI, and AutoGen.
Architect agent orchestration patterns, including planning, tool use, persistent state management, memory, reflection, and multi-agent coordination.
Develop and optimize RAG (Retrieval-Augmented Generation) pipelines, including document processing, chunking strategies, embedding workflows, and vector database integration.
Build robust agent evaluation, testing, and observability frameworks to ensure reliability and performance in production.
Design natural language-to-data query solutions integrating with platforms such as Databricks Genie.

Integrate and manage LLM/SLM services (OpenAI, Azure OpenAI, Anthropic, and open-source models), selecting appropriate models and applying prompt engineering and cost optimization.
Design prompt engineering strategies, including chain-of-thought, few-shot, and structured-output techniques for reliable agent behavior.
Implement guardrails, safety mechanisms, and content filtering for AI-generated outputs.
Evaluate and benchmark models for latency, accuracy, cost, and domain-specific performance.

Build scalable Python backend services with FastAPI to serve AI agent workflows to production applications at enterprise scale.
Design and implement caching, rate limiting, persistent agent state, and conversation memory strategies.
Develop event-driven microservices and real-time streaming for AI agent interactions.
Develop APIs and integration layers that connect AI agents with enterprise data sources, tools, and external services.
Implement distributed task processing with Celery and event-driven autoscaling with KEDA for production AI workloads.

Stay current with the rapidly evolving agentic AI landscape and evaluate emerging frameworks, models, and techniques.
Lead proof-of-concept development for new AI capabilities, moving successful experiments into production.
Mentor engineers on AI engineering best practices, prompt engineering, and agent design patterns.
Contribute to technical documentation, architecture decision records, and AI solution design specifications.
Champion the adoption of AI-powered development tools, such as Cursor AI and GitHub Copilot, across engineering teams.

Strong proficiency in Python, with hands-on experience building production AI applications.
Demonstrated experience with LangGraph or similar agentic AI frameworks (LangChain, CrewAI, or AutoGen) in production systems.
Hands-on experience with LLM API integration (OpenAI, Azure OpenAI, or Anthropic) and prompt engineering.
Experience designing and implementing RAG systems, including embedding models, vector databases, and retrieval strategies.
Solid understanding of multi-agent system design, agent orchestration, persistent state management, and memory patterns.
Experience with Python web frameworks (FastAPI) and distributed task processing (Celery) for production APIs.
Experience with event-driven microservices and real-time streaming patterns.
Proficiency with AI-powered development tools (Cursor AI, GitHub Copilot, or similar) for AI-augmented software development across the software development lifecycle.
Proficiency with Git, CI/CD pipelines, and cloud platforms, preferably Azure.

Experience with vector databases (Qdrant, Pinecone, Weaviate, or ChromaDB).
Experience with Databricks Genie or similar natural language-to-data query platforms.
Experience with AWS Bedrock AgentCore for managed agent runtime and multi-cloud agent deployment.
Knowledge of model fine-tuning, quantization, and serving optimization.
Experience with multi-tenant architecture patterns and enterprise-scale AI systems.
Experience with containerization (Docker, Kubernetes) and event-driven autoscaling (KEDA).
Understanding of AI safety, responsible AI principles, and enterprise governance requirements.

Primary framework: LangGraph (multi-agent orchestration with persistent state)
Additional frameworks: LangChain, CrewAI, AutoGen
LLM providers: OpenAI (e.g., GPT models), Azure OpenAI, Anthropic (Claude), and enterprise LLM services
Techniques: RAG, prompt engineering, chain-of-thought, function calling, and structured outputs
Data intelligence: Databricks Genie (natural language to SQL)
Vector databases: Qdrant, Pinecone, Weaviate, ChromaDB
Multi-cloud: AWS Bedrock AgentCore (managed agent runtime)
Patterns: Multi-agent orchestration, tool use, persistent state, memory management, and agent evaluation

Bachelor’s degree in Computer Science, Engineering, AI/ML, or a related technical field, or equivalent professional experience.
7+ years of software engineering experience, including significant hands-on AI/ML work in enterprise environments.
Strong communication skills, with the ability to explain complex AI concepts to technical and non-technical stakeholders.
Strong knowledge of Agile methodologies and principles.
Demonstrated interest in staying current with the rapidly evolving AI landscape.