AI Engineer

About the Role

We are seeking an experienced AI Engineer specializing in Agentic AI to design, build, and deploy autonomous AI agent systems on AWS. You will apply the AI-DLC (AI Development Life Cycle) framework to systematically develop AI solutions — from ideation to production — and architect intelligent agents that can reason, plan, use tools, and take actions to accomplish complex tasks. Leveraging Amazon Bedrock Agents, Amazon AgentCore, and related AWS services, you will build MVPs from scratch and iterate them into production-grade autonomous systems.

Key Responsibilities

Agentic AI Design & Development

1. Problem Framing — define business problems as agentic AI opportunities

2. Data & Knowledge Preparation — curate knowledge bases, design data pipelines 3. Architecture Design — select agent patterns, models, and AWS services

4. Rapid Prototyping & MVP — build functional MVPs from scratch to validate concepts quickly

5. Evaluation & Testing — benchmark agent performance, accuracy, safety, and cost

6. Deployment & Scaling — productionize on AWS infrastructure

7. Monitoring & Iteration — observe, learn, and continuously improve agent behavior

• Build MVPs from scratch using the AI-DLC framework — rapidly prototype agentic AI solutions to demonstrate feasibility and business value before scaling

• Establish and evangelize AI-DLC best practices within the team

• Architect multi-agent systems with agent collaboration, delegation, and orchestration patterns

• Develop agentic workflows using Amazon Bedrock Agents — including action groups, knowledge base integration, guardrails, and agent chaining

• Deploy and manage agents at scale using Amazon AgentCore for runtime orchestration, observability, and lifecycle management

• Implement agentic patterns: ReAct, Plan-and-Execute, Chain-of-Thought, Reflection, and Human-in-the-Loop

• Design and build tool integrations (APIs, databases, external services) that agents can invoke autonomously

• Build memory systems for agents — short-term (conversational), long-term (persistent), and episodic memory

RAG & Knowledge Systems

• Build production-grade RAG (Retrieval-Augmented Generation) pipelines as knowledge backends for agents • Design document ingestion pipelines: chunking strategies, embedding model selection, metadata enrichment

• Build and manage Amazon Bedrock Knowledge Bases (managed RAG service) for agent grounding and fact retrieval

• Implement and optimize vector store backends for RAG: – Amazon S3 with Vector Index — serverless vector storage integrated with Bedrock Knowledge Bases – PostgreSQL (pgvector) — via Amazon RDS / Aurora PostgreSQL for hybrid relational + vector workloads – Amazon OpenSearch Serverless — scalable vector search with filtering and hybrid retrieval – Pinecone — managed vector database for high-performance similarity search at scale

• Design retrieval strategies: semantic search, hybrid search (keyword + vector), metadata filtering, reranking, and query transformation

• Optimize RAG pipelines for accuracy, latency, and cost — including embedding model selection (Titan, Cohere), chunk size tuning, and retrieval evaluation

AWS Infrastructure & Production

• Architect agent infrastructure on AWS compute: EC2 (GPU instances), AWS Lambda (serverless inference), Amazon ECS / EKS (containerized agent services)

• Design data layers using Amazon RDS, DynamoDB, Aurora, and S3 for agent state, memory, and knowledge storage

• Implement agent observability: logging, tracing, cost monitoring, and performance metrics

• Build CI/CD pipelines for agent deployment and version management

• Ensure security, compliance, and responsible AI practices (guardrails, content filtering, PII handling)

Innovation & Collaboration

• Stay current with the latest agentic AI research (tool use, reasoning, planning, multiagent coordination)

• Evaluate and integrate new AWS AI service releases into agent architectures

• Collaborate with product, engineering, and business teams to identify and deliver agentic AI use cases

• Prototype novel agent capabilities and demonstrate business value through POCs

Required Qualifications

• Education: Bachelor's or Master's degree in Computer Science, AI, Data Science, or related field

• Experience: 2+ years in AI/ML engineering, with demonstrated experience building AI agent systems

• Strong programming skills in Python; familiarity with TypeScript/Node.js is a plus

Deep understanding of Agentic AI concepts:

– Agent architectures (single-agent, multi-agent, hierarchical)

– Reasoning & planning (ReAct, CoT, Tree-of-Thought, Plan-and-Execute)

– AI-DLC framework — structured methodology for AI solution development lifecycle

– Rapid MVP development

— ability to go from concept to working prototype quickly

– Tool use & function calling – Memory & state management

– Guardrails & safety for autonomous systems

Proven ability to build AI MVPs from scratch — take a business problem, apply AI-DLC methodology, and deliver a working agentic prototype within days/weeks.

Hands-on experience with AWS AI/ML services:

– Amazon Bedrock

— Agents, Knowledge Bases, Guardrails, foundation model APIs

– Amazon AgentCore

— agent deployment, orchestration, and runtime management

– Amazon SageMaker

— model training, fine-tuning, endpoints

AWS compute & infrastructure:

– EC2 — GPU instances (P4/P5, G5/G6) for model inference

– AWS Lambda

— serverless functions for agent actions and event-driven triggers

– Amazon ECS / EKS

— containerized agent microservices

Database & storage:

– Amazon RDS / Aurora

— relational data, pgvector for embeddings

– DynamoDB

— agent state and session storage – Amazon OpenSearch

— vector search for RAG

– Amazon S3

— document storage, model artifacts

Hands-on experience with vector stores for RAG:

– Amazon Bedrock Knowledge Bases

— managed RAG with automated ingestion and sync

– Amazon S3 Vector Index

— serverless vector storage – PostgreSQL pgvector

— vector extensions on RDS/Aurora PostgreSQL

– Amazon OpenSearch Serverless — vector collections with hybrid search

– Pinecone

— managed vector DB for production RAG

• Experience with agent frameworks: LangChain/LangGraph, LlamaIndex, CrewAI, AutoGen, or equivalent

• Understanding of embedding models (Amazon Titan Embeddings, Cohere Embed, OpenAI Embeddings) and retrieval evaluation metrics

• Proficiency with Docker, container orchestration, and API design (REST/GraphQL)