I build production-minded B2B software, internal tools, and modern web applications using AI-native development workflows.

After 3.5+ years as an SDR selling AI and software solutions to teams around the world, I made the move from selling software to building it. In June 2025, I began teaching myself to code by shipping real products, studying production codebases, and solving increasingly complex technical problems in public.

Today, I use tools such as Claude Code, Lovable, and GitHub to move quickly from idea to working product. But AI is not a substitute for ownership.

I use AI to accelerate implementation, explore architecture, debug faster, write tests, find edge cases, and iterate on product decisions. I still own the product direction, system design, code review, security decisions, user experience, and the quality of what gets deployed.

AI gives me leverage. Product judgment, technical reasoning, and shipping discipline are still the job.

My sales background continues to shape how I build: I start with customer problems, focus on clear outcomes, and think about how a product will actually be sold, adopted, and maintained.

• B2B SaaS products and workflow software
• AI-assisted applications with human-reviewable outputs
• Developer tools and GitHub-native workflow automation
• Progressive web applications with modern, responsive UI
• Internal tools, dashboards, and operational systems
• Fast MVPs that can evolve into durable product foundations

I am especially interested in products at the intersection of:

• AI agents and developer workflows
• Compliance, governance, and operational readiness
• Policy-as-code and repository automation
• B2B workflow software
• AI-assisted product development
• Rapid validation of niche SaaS ideas

AI-native development is not just “prompting an app into existence.” My workflow combines rapid prototyping with code-level ownership.

• Build initial product surfaces and user flows quickly
• Move projects into GitHub-backed, local development workflows
• Use Claude Code to inspect unfamiliar code, trace bugs, plan changes, and implement scoped features
• Review generated code rather than treating it as a black box
• Define edge cases, error states, permissions, validation, and security boundaries
• Test flows manually and iteratively harden products before sharing them with users
• Use Git and pull-request-style workflows to preserve a maintainable codebase

I am particularly efficient when using Claude Code as a high-context engineering partner: breaking work into smaller implementation plans, understanding existing architecture, identifying failure modes, and moving from a polished interface to a more reliable application.

AI-assisted compliance-readiness software for growing SaaS teams preparing for SOC 2, ISO 27001, and related frameworks.

AuditReady helps teams turn scattered policies, evidence, and business documentation into a clearer, more actionable readiness workflow. The goal is not to replace auditors, lawyers, or security professionals; it is to make the first pass of compliance preparation more structured, traceable, and efficient.

What I built and worked through:

• Compliance-readiness workflows for policies, controls, owners, tasks, and evidence
• AI-assisted document analysis and gap identification
• Structured findings designed for human review
• Secure authentication, document handling, and multi-step product flows
• Edge-case analysis, bug fixing, and application hardening with Claude Code
• GitHub-synced workflow: rapid initial build in Lovable, then local development and refinement in VS Code / Claude Code

Stack: React, TypeScript, Vite, Tailwind CSS, shadcn/ui, Supabase, Supabase Auth, database/storage workflows, AI-assisted analysis

Focus: Compliance operations, audit preparation, document intelligence, B2B SaaS workflows

Policy-as-code for AI coding agents.

ContextOps is a GitHub-native control layer for the instructions AI coding agents use across software repositories. Engineering teams increasingly rely on files such as AGENTS.md, CLAUDE.md, GitHub Copilot instructions, Cursor rules, and related configuration files to guide how AI works in their codebases.

The problem: those instructions drift, become inconsistent across repositories, and are difficult to manage safely at scale.

ContextOps gives those instructions a versioned, reviewable, repository-aware lifecycle.

Define engineering policy once, validate it against repository-specific instructions, and ship changes through reviewable pull requests.

Core concepts:

• Central policy management for AI-agent instructions
• Repository-specific configuration and assignments
• Versioning and approval-oriented change management
• Validation and drift detection across repositories
• GitHub App integration for controlled repository access
• Reviewable diffs and pull-request-based changes
• Audit logs and tenant-aware operational records
• A fail-closed approach to ambiguous or risky sync states

Build approach:

• Designed and prototyped the product control plane in Lovable
• Synced the codebase to GitHub early
• Moved into VS Code and Claude Code for backend integration, GitHub App architecture, security testing, and edge-case handling
• Focused on building the GitHub execution layer as tested, reviewable code rather than relying solely on prompt-generated functionality

Stack: React, TypeScript, Tailwind CSS, shadcn/ui, Supabase, GitHub App architecture, GitHub API, webhook and pull-request workflows

Focus: Developer experience, AI governance, repository automation, policy-as-code, secure GitHub integrations

AI-powered product specification and PRD generation.

SpecMirror turns rough product ideas into clearer technical specifications, implementation plans, and product requirements documents. It is designed to reduce the gap between an early concept and the structured documentation needed to build it well.

Stack: React, TypeScript, Vite, Tailwind CSS, shadcn/ui, Framer Motion, Supabase, AI workflows, encrypted data handling

Focus: Product discovery, requirements clarity, product-to-engineering translation

A mobile-first micro-learning experience for people learning technology.

Spark explores how short, interactive lessons can make technical concepts more accessible and engaging for self-directed learners.

Stack: React Native, Expo, TypeScript, Tailwind CSS, shadcn/ui, Supabase

Focus: Mobile learning, gamification, accessible technical education

A scroll-driven product concept for a rugged tactical smartwatch.

Astra Prototype One is an exploratory product experience that combines industrial design, terrain intelligence, biometrics, and high-performance interface storytelling. It is a visual and technical exercise in building immersive, interactive product marketing experiences.

Stack: React, TypeScript, Vite, Tailwind CSS, shadcn/ui, Framer Motion

Live Concept: astraprototype.one

Focus: Interactive storytelling, premium UI, motion design, product visualization

I care about more than making an interface look polished. I think through the user, the workflow, the business model, the implementation constraints, and what needs to be true for a product to earn trust.

I use Claude Code and other AI tools as force multipliers. They help me learn faster and execute faster, but I stay responsible for understanding the systems I deploy.

My SDR background means I naturally think about customer pain, messaging, differentiation, objections, and outcomes. I build with an awareness that software needs to solve a real problem and be explainable to the people buying it.

I learn by building. I would rather ship a focused, testable version, collect feedback, find weaknesses, and improve it than spend months trying to design a perfect product in isolation.

• Advanced React, Python, and TypeScript patterns
• Secure GitHub App and webhook architecture
• AI-agent orchestration and context management
• Production-grade authentication, authorization, and multi-tenancy
• B2B SaaS validation and go-to-market systems
• Compliance and security workflow design
• Better testing, debugging, and codebase-hardening practices with Claude Code

X • LinkedIn