AI/ML Engineer · Product · Founder
I don’t describe AI skill— I ship it.
Four production systems built solo—a lead-gen SaaS, an adaptive K-12 tutor, an approval-first email product, and a compliance-grade fintech ledger. This site is itself one of them.
- TypeScript
- Python
- Next.js
- RAG
- PostgreSQL
- Render
The work
Four shipped systems. Each leads with the outcome.
Tap any card to descend into the architecture, the model and stack rationale, and an honest tradeoffs section.
- AI SaaS · Lead Generation
LeadForager
3-phase verify · search → enrich → CRM
Full lead-gen SaaS: discover local businesses, verify them through a hybrid three-phase engine, AI-enrich, and push straight into a CRM.
Next.jsTypeScriptOpenAI / Anthropic / DeepSeekPuppeteerStripeHubSpot / Salesforce - Adaptive AI · EdTech
Agent Tutor K-12
1 concept, 5 ways · 0 curriculum stored
A learning-style-adaptive tutor that teaches the same concept five ways — storing only how a child learns, never what, with Socratic guardrails the personalization can't override.
PineconeRAGPrompt ArchitectureTypeScriptCOPPA-aware - Production SaaS · Approval-First AI
AI Auto Email Responder
Shipped · paying customers · never auto-sends
A live commercial SaaS that drafts your email replies over IMAP but never sends without your sign-off — from ingestion to Stripe billing, built to production solo.
Redis + BullIMAPStreaming LLMsStripeCross-instance rate limiting - Regulated Fintech · Compliance
SGO Compliance Ledger
Every dollar hash-chained · reconciled daily
A production donation & scholarship ledger where every dollar is hash-chained, double-entered, and reconciled against the bank daily — because 'probably right' isn't good enough for the IRS.
Anonymized client engagement — synthetic dataPostgreSQLDouble-entry ledgerSHA-256 hash chainSERIALIZABLEStripe webhooksNUMERIC(19,4)
The stack
What I build with.
Every tool here is load-bearing in a shipped system above—no aspirational logos.
| Languages | TypeScriptPythonSQL |
|---|---|
| AI / LLM | OpenAIAnthropicDeepSeekRAGPineconeStreaming LLMs |
| Frameworks | Next.jsReactNode.jsFastAPI |
| Data / Infra | PostgreSQLRedis + BullPuppeteer |
| Cloud & Deploy | RenderStripeCI/CD |
| Practices | Approval-first AIRate limitingEval suitesCost ceilingsDouble-entry correctness |
Hiring for AI/ML, product, or a founding role?
The fastest path is a direct one. I reply personally.