Berkeley Haas MBA
Graduated May 2026. AI for Business certificate, first cohort. Building Trackly through the MBA job-search problem.
Kevin
Astuhuaman
AI Product Manager
Berkeley Haas MBA, May 2026
Bay Area
Candidate thesis
I build AI products that move from prototype to production.
My strongest fit is AI product work where the proof has to be technical and operational: PayPal AI observability across 15+ checkout systems, digital-first financial products tied to 100K+ SMBs at BCP/Credicorp, and Trackly, a real-time job-search agent with web, iOS, macOS, CLI, and MCP surfaces.
Builder packet
The portfolio is the evidence system.
Trackly is the flagship artifact: a live product monitoring 1,900+ companies and 128,000+ roles, plus native apps, a CLI, an MCP server, and the operational stack behind it. The work below is organized by what it demonstrates: product judgment, systems thinking, and shipping.
Machine-readable profile
Structured context for recruiters, agents, and search.
This site exposes the same canonical facts as HTML, Markdown, JSON, and agent-skill metadata: llms.txt, profile.json, proof.json, and agent-skills.
15+
PayPal checkout systems touched by AI observability tooling.
1,900+
Companies monitored by Trackly from direct career pages.
100K+
SMBs reached through digital-first financial products at BCP/Credicorp.
500+
Trackly monthly active users from the latest internal snapshot.
AI product operator dossier
Recruiter packet
Everything needed to evaluate fit in one pass.
A compact set of human-readable and machine-readable artifacts for hiring managers, recruiters, and AI agents checking Kevin's background before a call.
01
One-page packet
Target roles, strongest proof points, work authorization, and links in one recruiter scan.
02 Resume snapshotReadable HTML and Markdown resume copy with public-safe PayPal, BCP, and Trackly wording.
03 Proof ledgerPublic proof, resume-approved claims, internal snapshots, and claims to avoid overusing.
04 Agent profileCanonical `llms.txt`, JSON profile, proof JSON, and agent-skill metadata for AI systems.
Selected Work
Four artifacts, each chosen because it proves a different part of the hiring thesis.
Trackly
Real-time recruiting agent that scans career pages directly and surfaces roles before job boards. Built across consumer UX, native apps, backend monitors, and agent integrations.
PayPal AI Observability
Internship project focused on AI observability tooling for checkout systems, reducing issue detection time by 75% in the project workflow.
SMB Fintech Products
Digital-first lending, insurance, and payments work tied to a $620M ARR/P&L portfolio and expanded financial access for 100K+ small businesses and underserved customers.
Risk
Credit
Insurance
SMB CRM
Digital onboarding
Payments
Agentic Development Workflows
Public GitHub work spanning MCP servers, LLM observability, evaluation pipelines, and AI-assisted shipping systems.
MCP
Evals
PostHog
Langfuse
SwiftUI
TypeScript
History
A tight timeline for the recruiter scan.
PayPal Checkout
Technical PM intern on AI observability tooling for checkout diagnostics and product-team workflows.
BCP / Credicorp
Product leadership across digital-first financial products for small businesses and underserved customers.
Builder and operator
Public work across Trackly, MCP, LLM observability, evaluation frameworks, and content distribution.
Proof Ledger
Every strong claim is labeled by evidence strength to avoid stale or overclaimed public copy.
Trackly coverage
1,900+ companies and 128,975+ roles appear on the live Trackly public site.
Public URL
Trackly App Store
iPhone, iPad, and Mac listing for Trackly: Apply first.
Public URL
Berkeley MBA and BCP background
PUCP public profile covers Berkeley admission, BCP role, and 100K+ access-to-credit impact.
Public URL
PayPal metrics
Use resume-approved and interview-safe wording. Do not expose confidential intern artifacts.
Proof notes