Ben Kehoe Portfolio
AWS Serverless Hero maintaining a set of widely-used open-source tools (aws-assume-role-lib, aws-export-credentials, aws-error-utils) that solve real cross-account IAM and credential-handling problems encountered while architecting serverless systems at scale.
Location
Coordinates unknown. Vibes: international.
Skills
What makes it work
A breakdown of the choices that make this portfolio stand out.
Small, single-purpose tools that fix a specific operational pain point
Each repository (assume-role chaining, credential export, error handling) solves one concrete AWS IAM annoyance that shows up repeatedly in serverless architecture work, rather than proposing a general-purpose framework.
A day job and title stated plainly alongside the open-source work
Listing the current employer directly on the profile grounds the tooling in ongoing, practical use rather than presenting it as abandoned side projects.
Adoption signals left visible rather than asserted
Star counts, forks, and GitHub achievement badges are left as public, checkable numbers rather than a claimed "widely used" line — the evidence is inspectable rather than asserted.
What Cloud Architects can take from this
Specific, actionable tips to apply to your own portfolio — no generic advice.
Ship small, single-purpose tools that fix one specific pain point in your domain — focused utility is easier for others to adopt and evaluate than a broad framework.
Keep your current role visible next to your open-source work so visitors can see the tools are informed by ongoing, real production use.
Let adoption metrics (stars, forks, issues closed) speak for themselves instead of writing marketing claims about your tools' popularity.
Projects
Three small Python libraries, each fixing one piece of boto3 friction: role assumption, credential export and error handling. Ben Kehoe's GitHub shows what an IAM-focused architect automates after hitting the same problem too many times.
A library that collapses boto3's manual AssumeRole boilerplate into one call that returns a session. The session refreshes expired credentials automatically and generates a role session name if you don't pass one. In Lambda, the session can be created once at init instead of calling AssumeRole on every invocation.
Cross-account role chaining is a common source of production bugs in multi-account AWS setups. The library fixes expiry handling in one place.
One-line assume_role(session, role_arn) instead of manual STS plumbing
Automatic credential refresh; no per-invocation AssumeRole in Lambda
161 stars, 12 forks
A CLI that takes credentials from any AWS profile, including AWS SSO, and passes them to other programs. It can print credential_process-compatible JSON, emit env vars, exec a command with credentials injected, or write to ~/.aws/credentials. It uses boto3's own credential resolution instead of scraping the CLI's cache files.
AWS later shipped its core behavior natively in the CLI, so he identified a real gap early. The README says so plainly.
Works with any profile type boto3 supports, including AWS SSO
Outputs: credential_process JSON, env vars, exec wrapper, credentials file
Overlapping feature later added to the AWS CLI; README points users to it
Helpers that make botocore's generic ClientError easier to handle. Instead of reading error.response['Error']['Code'] and re-raising, you catch errors.NoSuchBucket directly, or use catch_aws_error('NoSuchBucket'), and read error.message.
Readable error handling around AWS calls is what keeps production code maintainable. This removes a common source of boilerplate and missed error codes.
except errors.NoSuchBucket instead of string-matching error codes
Single-file option, no extra dependencies beyond botocore
151 stars, 8 forks
More Cloud Architect portfolios
Ready to build your portfolio?
Follow the Cloud Architect roadmap — skills, projects, and timeline to get hired.