AI & Cloud Engineer

Agentic AI workflows,
engineered for production.

I design and ship production-grade AI systems — multi-agent orchestration with LangGraph, retrieval pipelines with LangChain, and cloud-native deployments on AWS.

View Projects Get in Touch

About

Engineering AI that ships

I'm an engineer focused on the intersection of large language models, agentic architectures, and cloud infrastructure. I turn research concepts into reliable systems that teams can operate in production.

My work spans graph-based agent orchestration, RAG pipelines, event-driven workflows, and infrastructure-as-code on AWS — always with observability, security, and cost efficiency in mind.

12+

Agent workflows shipped

25+

AWS services integrated

Production LLM pipelines

Expertise

Skills & Technologies

AI & Agents

LangGraph
LangChain
Multi-agent orchestration
RAG & vector search
Prompt engineering
Bedrock & OpenAI

AWS Cloud

ECS Fargate
Lambda
S3 & CloudFront
EventBridge
Step Functions
Bedrock
OpenSearch
CDK / IaC

Engineering

TypeScript
Python
Next.js
Docker
CI/CD
Observability
API design

Portfolio

Agentic Workflow Projects

Production systems built with LangGraph, LangChain, and AWS — from multi-agent orchestration to event-driven document pipelines.

Multi-Agent Research Orchestrator

A LangGraph-powered research assistant that decomposes complex queries into parallel sub-tasks, synthesizes findings, and cites sources — deployed on AWS with full tracing.

Supervisor + worker agent graph with conditional routing
Bedrock Claude for reasoning, Titan for embeddings
DynamoDB session state with TTL

LangGraphBedrockDynamoDBLambdaCDK

API Gateway → Lambda (graph runner) → Bedrock · DynamoDB · S3

Enterprise RAG Knowledge Pipeline

End-to-end document ingestion and retrieval system using LangChain, with hybrid search across OpenSearch and semantic re-ranking for high-precision answers.

S3 event-driven ingestion with Step Functions
Chunking strategies tuned per document type
Evaluation harness with RAGAS metrics

LangChainOpenSearchS3Step FunctionsSQS

S3 → EventBridge → Step Functions → Lambda · OpenSearch

Agentic DevOps Assistant

An autonomous workflow agent that triages CI failures, inspects logs, proposes fixes, and opens pull requests — orchestrated as a stateful LangGraph with human-in-the-loop gates.

Tool-calling agents for GitHub, CloudWatch, and ECS
Human approval nodes before destructive actions
CloudWatch dashboards for agent run metrics

LangGraphGitHub APIECSCloudWatchSNS

EventBridge → ECS Fargate (agent) → GitHub · CloudWatch

Document Intelligence Workflow

Serverless pipeline that extracts structured data from invoices and contracts using vision-capable models, with confidence scoring and exception queues for review.

Multi-modal Bedrock models for OCR + extraction
Dead-letter queues for low-confidence outputs
Aurora for structured output storage

BedrockLambdaSQSAuroraEventBridge

S3 upload → SQS → Lambda (extract) → Aurora · SNS alerts

Infrastructure

Production AWS Architecture

This portfolio is designed for containerized deployment on AWS using infrastructure-as-code. The reference stack provisions ECR, ECS Fargate behind an Application Load Balancer, with optional CloudFront for global edge caching.

Amazon ECR

Container image registry for the Next.js standalone build

ECS Fargate

Serverless compute running the containerized app

Application Load Balancer

HTTPS termination, health checks, and traffic routing

CloudFront (optional)

Global CDN for static assets and edge caching

AWS CDK

Infrastructure as code — reproducible, version-controlled stacks

Internet → CloudFront → ALB → ECS Fargate (Next.js)
                              ↓
                         ECR ← CI/CD pipeline
                              ↓
                    CDK Stack (IaC)

Contact

Let's build something intelligent

Open to collaborations on agentic AI systems, cloud architecture, and production LLM workflows.

Send Email LinkedIn GitHub

brian@example.com

Agentic AI workflows,engineered for production.