Principal Software Engineer @ PayPal

Shipping AI‑enabled software delivery at enterprise scale.

I architect the systems that let 10,000+ PayPal engineers build with AI: autonomous agent pipelines, the company's largest internal MCP server, and the observability that keeps it all honest. Ten years, MTS I to Principal, one motto: automate everything.

mike@baltimore: zsh
whoami --verbose
name: Michael N. Bukosky
role: Principal Software Engineer, PayPal
focus: AI-SDLC · MCP · Observability
mcp_users: 1600+ # 2025 peak
engineers_enabled: 10000+
savings_driven: multi-$M # 3yr est.
awards: Miro Hall of Fame ’26 · CTO Innovation ’25 · Copilot Challenge ’25
1,600+
engineers on my MCP server at its 2025 peak
10,000+
engineers tracked by my Claude Code adoption framework
multi-$M
est. 3-year savings from the SignalFx → Datadog migration
3,000+
services at ~95% tracing coverage across Core, Venmo & Braintree
pipeline

The platform that ships the software.

A live map of PayPal's autonomous AI-SDLC platform: a work item enters at the left, a planner drafts against the repo, agents implement in parallel, and nothing reaches a draft PR until every gate passes. A human still owns the merge.

ai-sdlc.control-plane
topology 13 nodes in-flight merged 0
Pipeline diagram unavailable — the stages are listed below.

hover or tab a node for stage detail
  1. Work item intake — Jira and GitHub issues normalized into agent-ready specs.
  2. Triage & routing — classify, size, and route by ownership.
  3. Planning agent — Claude Code drafts an implementation plan against the repo.
  4. Context assembly — MCP connectors pull ownership, docs, and prior art.
  5. Implementation agents — parallel Claude Code workers write the change.
  6. Test gate — unit and isolated component tests with WireMock.
  7. Static analysis — SonarQube quality gate.
  8. Security gate — dependency and secret scanning before any PR opens.
  9. Draft pull request — opened only after every gate passes.
  10. Human review — an engineer owns the merge decision.
  11. Merge and deploy — DORA metrics fed back into intake.
ai-work

Tools teams ship with.

Developer-experience work at PayPal: the AI-SDLC platform, MCP servers, and governance that engineering teams build and ship on — behind auth, quality gates, and the guardrails that let them move fast without breaking production.

paypal/ai-sdlc-platform

Autonomous AI-SDLC Platform

Converts work items into Claude Code agents that run full implementation workflows behind quality gates and open draft PRs for human review. I'm the architect and lead engineer, and the author of PayPal's AI-enabled SDLC vision combining agents, review gates, and DORA metrics.

✓ new teams onboard in < 10 mins
GKELiteLLMOTelClaude CodeDORA
paypal/smb-fs-mcp-server

SMB-FS MCP Server

PayPal's largest internal MCP server at its 2025 peak, built on a first-at-PayPal PingID OAuth flow, Datadog observability, and RBAC via MyAccess. Shipped Slack support before an official MCP existed, then migrated teams to first-party vendor servers as they arrived; today it serves internal-only integrations like Ownership and SonarQube. The security pattern it defined is still the one internal MCP servers follow.

✓ 1,600+ users at 2025 peak · pattern still the standard
MCPPingID OAuthRBACOwnershipSonarQube
paypal/ai-vendor-partnerships

AI Vendor Partnerships

PayPal's bridge to the AI vendor ecosystem: design partner with Anthropic, OpenAI, Cursor, Miro, WireMock, and Datadog, evaluating emerging capabilities early and translating them into production developer enablement for 10,000+ engineers.

✓ design partner to 6 AI vendors
AnthropicOpenAICursorMiroWireMockDatadog
paypal/ai-hub

PayPal AI Hub & MCP Ecosystem

I govern PayPal's official AI Hub, setting the standards for Skills, Plugins, and MCP connectors and reviewing every marketplace submission. I also pioneered the GitHub, Jira/Confluence, Datadog, and Slack MCP integrations behind AI-assisted code review and incident investigation.

✓ 29,000+ doc views · 4,500+ engineers
AI governanceMCP connectorsDevRel
career

git log --author="bukosky"

a1f9e2cDec 2025 – present

Principal Software Engineer · PayPal

Architect and lead engineer of PayPal's autonomous AI-SDLC platform, which turns work items into Claude Code agents that run full implementation workflows behind quality gates and open draft PRs for review, with new teams onboarding in under ten minutes. I own PayPal's AI Hub and set the standards for its Skills, Plugins, and MCP connectors, and I'm the company's design-partner bridge to Anthropic, OpenAI, Cursor, Miro, WireMock, and Datadog.

8c47bd1Feb 2023 – Dec 2025

Senior Staff Software Engineer · PayPal

Established AI in the SDLC. Built the SMB-FS MCP server on a first-at-PayPal PingID OAuth flow, growing it to over 1,600 users at its 2025 peak, and stood up an adoption framework tracking 10,000 engineers alongside an AI Bootcamp and Coffee Talks. Hand-picked by a VP to co-lead the Observability Guild, where I drove the enterprise SignalFx to Datadog migration and Universal Tracing to roughly 95 percent of 3,000+ services. CTO Innovation Excellence first runner-up and Copilot Challenge winner.

2b90c44Aug 2016 – Feb 2023

Member of Technical Staff I & II · PayPal

Migrated roughly 100 Credit apps onto enterprise CI/CD in a plan the CTO reviewed personally, and created Isolated Component Testing with WireMock, which grew into a company-wide standard. Built Terraform providers automating Raptor CI/CD, owned ActiveMQ Artemis with zero-downtime upgrades, and delivered the Pay-in-4 platform.

# earlier commits
2015–2016 Mind Over Machines · Senior Consultant, C#/Node.js microservices lead
2014–2015 HookLogic · Senior Platform Developer, retail search exchange
2008–2014 FactSet · Senior Software Engineer / Architect, content-collection at scale
2007 B.S. Computer Engineering · University of Maryland, College Park
skills

My happy place.

ai_and_dev_tooling

MCP, Claude / Claude Code, Codex, Gemini, GitHub Copilot, LiteLLM, AI governance, LLM adoption tooling

observability

Datadog, OpenTelemetry, Splunk, APM, distributed tracing, SLOs, DORA metrics

cloud_and_cicd

AWS, EKS, GKE, Kubernetes, Docker, Harness, Jenkins, Terraform

languages_and_frameworks

Java 8/17, Go, JavaScript / Node.js, C#/.NET, Spring Boot, GraphQL, ActiveMQ Artemis, CloudEvents

recognition

Awards and recognition.

Awards

  • Canvas '26 Hall of Fame: "The Amplifier"official vendor award · Miro · 2026
  • CTO Innovation Excellence Award, First Runner-upcompany-wide · PayPal · 2025
  • GitHub Copilot Challenge WinnerPayPal · 2025
  • Emerging Leadership Program graduate18-week program · 2022

Writing & talks

  • Internal AI-tooling guides29,000+ views · 4,500+ unique engineers
  • AI Coffee Talks · Office Hours · AI Bootcamprecurring internal enablement series
side-project

My fun zone.

Spotify Unchained

Archiving my Spotify Discover Weekly every week since 2015 — a decade of playlists, preserved before they roll over. A personal project I've quietly kept alive the whole time.

✓ running since 2015
side projectautomationspotify api

My Brick Locker

A digital LEGO collection tracker — inventory, piece counts, values, and wishlists. I built it with my son so he'd stop asking me how many LEGO sets he had and what they were worth. Now he can look it up himself.

✓ built with my kid
side projectlegoinventory