Full-stack software engineer building AI SaaS platforms, developer tools, eval workflows, and secure model/runtime infrastructure.
m0lz.dev · LinkedIn · Email · GitHub
I help teams turn AI prototypes, internal tools, and high-leverage product ideas into production SaaS systems: data contracts, model/runtime choices, secure infrastructure, evals, observability, deployment gates, and user-facing workflows that hold up after launch.
Most recently at REVGEN, I built production AI and SaaS systems across Next.js, React, TypeScript, Python, Supabase/PostgreSQL, Docker, RunPod, Vercel, GitHub Actions, Twilio, SendGrid, Stripe, QuickBooks, Gmail, and ShareFile.
I am currently focused on full-stack software engineering, AI SaaS platforms, contract engineering, applied AI consulting, developer tools, eval infrastructure, secure model runtimes, and platform/reliability work.
| Problem | What I build |
|---|---|
| AI feature stuck at prototype | Production workflows with data boundaries, model/runtime selection, evals, observability, and release gates |
| Code review or eval work for AI labs | Coding-task evaluation, edge-case design, regression tests, review rubrics, and failure taxonomy |
| Secure or private AI deployment | Dockerized inference, open-weight model runtime tradeoffs, closed-environment operation, audit trails, and PII controls |
| SaaS workflows that need engineering/product judgment | Multi-tenant apps, RLS/auth, integrations, feature flags, rollout plans, and customer-facing automation |
| Developer-tooling or platform gaps | CLIs, MCP tools, workflow orchestration, CI parity, release automation, metrics, and smoke-test evidence |
| OSS or existing-codebase fixes | Focused patches that fit local architecture, respond to maintainer feedback, and include targeted validation |
m0lz.01 is the local idea-to-distribution pipeline behind m0lz.dev: research, benchmark, draft, evaluate, publish, and distribute technical content.
m0lz.02 is a Rust/TypeScript AI coding workflow orchestrator: Plan -> Implement -> Contract-Evaluate, dual-model adversarial review, context-isolated evaluators, Docker CI parity, Windows smoke tests, npm distribution, and release evidence.
m0lz.03 is a TypeScript MCP security proxy: OAuth 2.1/API-key auth, permission scoping, rate limiting, bidirectional PII detection, queryable audit logs, optional SQLCipher, Docker packaging, and benchmark coverage.
Status snapshot checked with GitHub CLI on June 7, 2026.
| Lane | Project | Work item | Status |
|---|---|---|---|
| Active review | Puppeteer | #15075 page-level locale emulation across CDP and WebDriver BiDi | Open, mergeable; CLA and visible checks green |
| Active review | AWS IAM Policy Autopilot | #217 configurable policy-generation resource cutoff | Implementation complete from my side; open, mergeable after signed rebase, awaiting maintainer review |
| Active review | AWS IAM Policy Autopilot | #227 read-only MCP server guardrail for IAM policy mutation | Open, mergeable; issue #92 linked |
| Active review | Deno / dprint TypeScript plugin | #800 preserve Markdown hard breaks in JSDoc formatting for Deno issue #31831 | Open, mergeable upstream fix |
| Active review | TrustGraph | #977 reject invalid PDF decoder input for issue #949 | Open, mergeable; CLA and tests green |
| Merged | TrustGraph | #955, #954, #952, #948 exception handling, cache parsing, metric-label parsing, and WebSocket fixes | Merged |
| Merged | AWS IAM Policy Autopilot | #214 pre-commit hooks and local validation documentation | Merged |
| Merged | Harper | #3460 Rust grammar false-positive fix for built-in verb phrases | Merged |
Application stack: TypeScript, JavaScript, Python, Rust, React, Next.js, Node.js, FastAPI, Django, Laravel, HTML, CSS, Tailwind CSS, Vite.
AI/ML: open-weight models, secure inference, model runtime selection, RAG, embeddings, rerankers, pgvector, MCP, multi-agent workflows, LLM orchestration, evaluation pipelines, Whisper, pyannote, Hugging Face, PyTorch, Vercel AI SDK, OpenAI API, Claude/Anthropic API.
Infrastructure and delivery: Docker, Docker Compose, RunPod, AWS, Vercel, GitHub Actions, GHCR, CI/CD, release automation, Supabase, Neon/PostgreSQL, Redis, Drizzle ORM.
Reliability and security: observability, telemetry, health checks, audit logging, feature flags, progressive rollouts, OAuth, JWT, RBAC, RLS, PII detection, rate limiting, SQLCipher, Playwright, Jest, Vitest, pytest.
SalesForge AI (internal): call-intelligence and coaching platform. Ingests Coffee/Levitate recordings, runs Dockerized RunPod inference with Whisper large-v3 and pyannote, extracts deterministic metrics, submits budget-controlled Anthropic Batch evaluations, and surfaces cited coaching insights in dashboards and chat. Includes durable heartbeats, Slack alerts, completion catch-up, poll-only reconciliation, production DB verification, and RunPod/Docker deploy reconciliation.
Bloom: multi-tenant SaaS for landscaping and service businesses. Customer segmentation, campaigns, proposals, invoice workflows, routing/dispatch concepts, conversational AI assistance, org-scoped auth, RLS-backed data isolation, feature flags, structured logging, E2E/regression suites, and integrations across Stripe, Twilio, SendGrid, Gmail, QuickBooks, and Vercel Analytics.
Alpaka: value-chain intelligence software using Next.js, Python FastAPI, Docker Compose, pgvector, semantic search, reranking, groundedness scoring, multi-agent synthesis, open-weight Qwen runtimes, Vercel, and RunPod.
Ready Text Legal: invite-only legal client communication workspace with Supabase auth/RLS, case/contact management, consent tracking, SMS/MMS workflows, multilingual templates, approval gates, automation controls, audit archives, billing, and launch-readiness validation.
| Project | What it proves | Stack |
|---|---|---|
| m0lz.00 | Portfolio and technical writing hub at m0lz.dev | Next.js, TypeScript, MDX |
| m0lz.01 | Local idea-to-distribution pipeline for research, drafting, evaluation, and publishing | TypeScript, SQLite, CLI |
| m0lz.02 | Contract-driven AI coding workflow orchestration and release evidence | Rust, TypeScript, JSON-RPC |
| m0lz.03 | MCP security proxy, permissions, audit logs, benchmarked tool boundaries | TypeScript, Docker, SQLite |
| m0lz.04 | AI-powered legal case management for pro se litigants | JavaScript, Claude API |
| Investor Matchmaker | Investor-founder meeting scheduler with greedy bipartite matching and Excel I/O | Python |
Before moving full time into engineering, I spent 15 years in B2B SaaS leadership at companies including 24/7 Software and CleanAir.ai. That background shows up in how I build: I care about deployment, adoption, customer workflows, business value, and operational ownership.
- MBA - Nova Southeastern University
- BSBA, Entrepreneurship - University of Central Florida