Open to work

I build software
that runs.

Backend by title, full-stack in practice. Deep in AI agent work at BambooHR; building in the margins.

3
Projects
51
Endpoints
22
Terminal cmds
$swarm workspace create --name ws1
Creating workspace ws1...
Neon branch: ws1-main
API URL: api-ws1.dev.bokendell.com
App URL: ws1.dev.bokendell.com
Status: ready
$swarm db migrate run --app ws1
Applying 3 migrations...
Done in 1.2s
$

Query this portfolio with AI

Real Mastra agent calling the public MCP server. Tool-call timeline on the right shows actual per-call latency. 10 free runs / IP / day — for a deeper trace, see /mcp/try →

/mcp/try →
connected to bokendell.com/mcp · tools: 8 · model: anthropic
10 runs / ip / day
suggested

"what are bo's strongest three projects and why?"

Press Run, or edit the prompt and Run. Tool calls stream into the right panel as the agent reaches for them.

tool callsidle
press Run — each tool the agent calls will appear here with real latency.
B

Bagman

shipping

voice-driven golf scoring for groups playing multiple game types simultaneously. one mastra agent, twenty-two tools, ably realtime.

t

this site

shipping

the portfolio site, built on the case-study infrastructure it documents. ships the 2026-native primitives — llms.txt, agent-card, mcp — without the AI-slop visual signatures that usually come with them.

b

bk · the monorepo cli

shipping

the personal cli that runs my monorepo. workspace isolation, deterministic ports, per-branch neon databases, stable tunnel urls — composed from one typescript binary built on oclif.

Activity

GitHub contributions

No contribution data

LeetCode

Easy0
Medium0
Hard0

Tech stack

TypeScriptHonooRPCReact NativeExpoPostgreSQLDrizzle ORMFly.ioMastraOpenRouterMCPReactTailwindNext.js

Let's talk.

Open to new opportunities, collaborations, and interesting problems.

last shipped
recently — github graphql rate-limited, retrying in <1h
⌘Kterminal