I like building AI agents around messy human workflows. Most of my work starts with scattered docs, sales calls, spreadsheets, or operational chaos. I turn that into full-stack software, automations, and agent workflows that help people get work done with more clarity.
Currently building with AI to learn and explore what AI-native work actually looks like.
Legal document-operations software for claims workflows: upload messy document bundles, classify and split files, extract structured fields with citations, review low-confidence outputs, and generate claim artifacts.
AI CRM and relationship-work agent harness: managed agent sessions, CRM tools, files, memory, approvals, Telegram workflows, automations, browser tasks, and evaluator traces.
AI-enabled GTM pipeline systems for startups: scrape TAMs, enrich accounts, score fit, verify contacts, draft briefs, and hand qualified leads into outbound workflows.
Karpathy-inspired LLM wiki for company memory: immutable source files, an agent-maintained markdown knowledge base, index/log workflows, Notion action state, and reviewed agent updates.
Singapore
Building reviewed AI systems across legal docs, CRM, GTM, company memory, and order intake.
San Francisco
Mapped enterprise intake and prior-auth workflows, including review flows that reduced manual staffing needs from ~14 to ~3.
Singapore, San Francisco, Australia
Built AI GTM systems for 8 startups, generating $1M+ pipeline and $200K revenue.
Singapore
Second outbound hire for Singapore expansion; hit 200%+ quota and closed 22 logos.
Singapore, APAC
Built seed-to-Series A enterprise traction from $0 to $450K ARR and closed 9 super-app partnerships.
Wolfson College
Graduated top 10%; machine-learning sentencing dissertation awarded First Class.