Skip to content
Laith Al-Saadoon
About

About Laith Al-Saadoon

about
whoami
laith. principal ai engineer, aws developer tools. customers: the agents that build on aws, and the humans behind them.
cat ~/.plan
measure how long an agent can work alone before a human has to step in. then make that number bigger.
ls ~/building
symspec/ find contradictions in structured requirements with z3. microvms-agentd/ exec + file transfer for aws lambda microvms. rust, python, node. atif-sql/ query agent trajectories with sql. memhtml/ inspect and review agent memory as html in git.
cat ~/aws-work
tech lead agent toolkit for aws (github.com/aws/agent-toolkit-for-aws). creator aws mcp servers (github.com/awslabs/mcp).
grep keynote ~/history
2023 generative ai cdk constructs. werner vogels, re:invent. 2024 pagerduty + amazon q incident integration. matt garman, re:invent.
uptime
up 10 years at aws, 3 roles: solutions architect, principal prototyping architect, principal ai engineer. before that: one of the first ten at onica (later rackspace). before that: a genomics lab.
cat /etc/motd
houston since 2015, riverside before that. wife sylvia, cat copper. fishing, paintball, fiction. the home gym is a raspberry pi wired to a traffic light.
exit 0

I'm Laith (rhymes with "faith") Al-Saadoon, a Principal AI Engineer in AWS Developer Tools. I build tools that help AI agents work longer without human intervention, and make their work easier to inspect, verify, and improve.

At AWS, I help coding assistants build well on our platform. I work with customers and the teams building those assistants, turn what we learn into open-source tools and reference implementations, and bring the gaps back to engineering.

I'm the tech lead for Agent Toolkit for AWS and created the AWS MCP Servers. My own fleet is a daily test bed: I watch where agents need help, build tools to address those failures, and measure what changes.

How I got here

I've been at AWS since 2016: first as a solutions architect, then a principal prototyping architect, and now a principal AI engineer. I led AWS MCP strategy and built the Generative AI CDK Constructs shown in Werner Vogels' re:Invent 2023 keynote.

Other work includes the PagerDuty and Amazon Q incident integration in Matt Garman's re:Invent 2024 keynote, and United Airlines' "Every Flight Has a Story" notifications on Amazon Bedrock and Claude.

Before AWS, I was one of the first ten people at Onica, a consultancy Rackspace later acquired. Before that, I had a biomedical sciences degree and a bench in a genomics lab and a developmental biology lab, until the computers won.

Away from the keyboard

I grew up in Riverside, California, and have lived in Houston since 2015 with my wife Sylvia and our cat Copper. I fish, play paintball, and read fiction. My home gym runs on a Raspberry Pi wired to a traffic light that tells me when to rest and when to go.

In myself and the people I work with, I value authenticity, integrity, ownership, empathy, and listening well. Credentials matter less to me than how we treat people and what we do with what we learn.

The Frontier Field Notes are where the work gets written up: long-form essays and short side quests.

Get in touch

For a conversation about the work, reach me on LinkedIn.

Writing elsewhere

The short version, in a terminal
about
whoami
laith. principal ai engineer, aws developer tools. customers: the agents that build on aws, and the humans behind them.
cat ~/.plan
measure how long an agent can work alone before a human has to step in. then make that number bigger.
ls ~/building
symspec/ find contradictions in structured requirements with z3. microvms-agentd/ exec + file transfer for aws lambda microvms. rust, python, node. atif-sql/ query agent trajectories with sql. memhtml/ inspect and review agent memory as html in git.
cat ~/aws-work
tech lead agent toolkit for aws (github.com/aws/agent-toolkit-for-aws). creator aws mcp servers (github.com/awslabs/mcp).
grep keynote ~/history
2023 generative ai cdk constructs. werner vogels, re:invent. 2024 pagerduty + amazon q incident integration. matt garman, re:invent.
uptime
up 10 years at aws, 3 roles: solutions architect, principal prototyping architect, principal ai engineer. before that: one of the first ten at onica (later rackspace). before that: a genomics lab.
cat /etc/motd
houston since 2015, riverside before that. wife sylvia, cat copper. fishing, paintball, fiction. the home gym is a raspberry pi wired to a traffic light.
exit 0