
Technical PM + Full-Stack AI Developer
I help companies ship faster. Solo.
Available for consulting, retainers, and contract engagements.
Get in touchYears Product Management
Products Shipped as PM
Products Built and Deployed Solo
AI Agents Built
I started with a Computer Science degree from Florida Tech, then earned a Master's in Information Systems at Stevens Institute of Technology. My career began at SiriusXM, where I drove the integration of Pandora's backend systems for Artist Radio Stations.
At Spryte Labs, I led a team of engineers to build a no-code website builder that cut time-to-market by 10x and scaled the client base 5x. At Influence Mobile, I refactored the main app for cross-platform deployment.
Then came Blockdaemon: 3.5 years as Technical Product Manager for their cryptographic key management and blockchain infrastructure products. I architected the AWS Marketplace deployment, launched self-checkout, designed a sandbox environment, and automated validator workflows across 30+ blockchain protocols. Sales cycles shortened by 70%. Infrastructure costs dropped by 50%.
In late 2025, I built Runwell Systems. Not a dev shop. An AI infrastructure company. I designed a 32-agent orchestration system that captures every decision as a learning signal and improves itself with each deployment. After shipping 23 products as a PM in the industry, I built and deployed 8 of my own. One operator.
Years later, I rebuilt the same website builder I created at Spryte Labs with a full engineering team. This time, I built it solo using Claude Code, and now I deliver full production websites in under five minutes.
I help companies ship faster. Available for consulting and contract engagements.
English, French. Product School NYC. Stevens PM Certificate.
Nov 2021 - Mar 2025, Technical Product Manager
Blockdaemon is the leading blockchain infrastructure company, serving institutional clients with key management, staking, and wallet products.
blockdaemon.com →Cryptographic Key Management
Enterprise-grade MPC key management for institutional blockchain clients. Designed and shipped the AWS Marketplace deployment flow that shortened enterprise sales cycles by 70%. Launched a pay-as-you-go self-checkout model that boosted sales 20% within 6 months. Created a sandbox environment that reduced demo complexity by 90%, enabling prospects to evaluate the product independently before engaging sales.
Unified API and Wallet Onboarding
Unified the API and Wallet onboarding into a single product journey, reducing friction for institutional clients. Integrated with the Node Management Dashboard for centralized multi-protocol control. Coordinated cross-team requirements between wallet, compliance, and infrastructure teams.
Validator Node Deployment
Automated validator node deployment and staking workflows across 30+ blockchain protocols, including Ethereum, Solana, and Cosmos. Integrated all protocols into a unified Node Management Dashboard. Managed server provisioning and lifecycle automation for bare-metal infrastructure.
Migrated blockchain nodes from cloud to bare metal, cutting infrastructure costs by 50%.
Shorter sales cycle
Sales boost in 6 months
Demo complexity reduction
Blockchain protocols
Infrastructure cost cut
Refactored the main client-facing mobile application for cross-platform deployment, achieving a 50% faster release cycle. Managed product requirements and sprint execution across engineering teams. Developed the integration plan for crypto and Bitcoin rewards features.
Led a team of engineers to design and launch a no-code website builder for an ETF firm. Cut time-to-market by 10x and scaled client capacity 5x. Architected backend systems and CI/CD pipelines (Jenkins, Docker). Hands-on Vue.js and Python development. Maintained 100% uptime throughout the engagement.
Drove the integration of Pandora's backend into SiriusXM's app for Artist Radio Stations. Coordinated cross-team requirements across audio streaming, content management, and recommendation systems. Conducted user metrics analysis and UX optimization. Managed bug triage and release coordination.
After shipping 23 products as a PM in the industry, I built these myself. Production products, not demos. Solo, using Claude Code and a 32-agent AI infrastructure.
Full business management platform with website CMS, online ordering, appointment booking, inventory management, e-commerce, and six embedded AI agents. White-label, multi-tenant architecture powering 11 client deployments.
Built as a Turborepo monorepo with 30+ packages. Embedded AI agents: Concierge (customer service), Sales Scout (lead generation), Market Research (trend analysis), AI Advisor (business guidance), Creative Agent (content generation), Analytics Agent (performance insights).
White-label AI chatbot infrastructure deployed in four distinct configurations. Each variant is production-deployed and serving real users.
Concierge: scrapes and generates contextual AI chatbots for any business URL. Shopimate: AI shopping assistant with vector search across Shopify catalogs. Jarvis: Telegram bot for coding from phone via Claude Code on VPS. CEGAudit TaxBot: RAG-powered tax advisor with hybrid search and grounded citations.
Standalone AI shopping assistant for Shopify merchants. Extracted from the BIB platform into its own monorepo with dedicated landing page and demo.
Vector search with pgvector embeddings across product catalogs. Scrape cache with SQLite for instant repeat visits. Docker deployment with PostgreSQL and Redis. 9 vendored packages from BIB. Branded landing page with Noir Rose palette and interactive live demo.
A platform for building software autonomously. Describe what you want; the system architects, generates code, runs tests, and deploys it.
10-phase cognitive loop with specialized agents for architecture, code generation, testing, and deployment. On-demand development environments with pre-configured EC2 instances. State persistence: snapshot and restore entire development sessions. Multi-AZ infrastructure with 99.9% uptime target.
Unified creative infrastructure for generating marketing assets. Image, video, voice, lip sync, and full ad pipeline from brief to final render. Used for 2 e-commerce brand ad campaigns.
Image generation: Gemini Imagen and Nano Banana. Video generation: Kling, Veo 3.1, SkyReels, and Sora. Voice generation: VoiceBox (Kokoro + Qwen3, 67 voice packs, 9 languages, voice cloning) and ElevenLabs. Lip sync: Hedra and HeyGen. Video assembly and post-production: Remotion (React) and ffmpeg. Full pipeline from creative brief to rendered ad.
Field-agnostic web scraping framework with Claude-powered semantic data extraction. Cascading fallback architecture that adapts to any target.
Starts with fast static HTML extraction, falls back to headless browser rendering, then escalates to stealth browser with bot detection evasion (Patchright) when needed. Supports REST and GraphQL APIs, Google Places, and meta-search across 70+ engines via SearxNG. Schema-driven output with token-aware cost tracking. Powers the Concierge chatbot (business scraping), Shopimate (product catalog extraction), Lead Scout (prospect discovery), and Reputation Scout (review aggregation).
At Spryte Labs in 2019, I built a no-code website builder with a team. Years later, I rebuilt the same capability solo using Claude Code.
Production websites delivered in under five minutes. 30+ sites deployed through the Runwell WebStudio platform. Full branding, responsive design, i18n (English, French, Arabic with RTL), and Docker deployment included.
Invoice conversion tool for the Tunisian market. Converts invoices to government-compliant Excel and XML (TEIF) formats for the national electronic invoicing system.
Three input methods: camera capture, file upload, and manual entry. AI-powered data extraction using Google Gemini. Trilingual: French, Arabic (RTL), English.
PRDs, roadmaps, sprint management, and go-to-market strategy for technical products. AWS Marketplace listing, self-serve models, sandbox environments, enterprise sales enablement.
70% shorter sales cycle at Blockdaemon
Design and build autonomous multi-agent systems with role-based delegation, token budgets, and self-improving learning loops. Hierarchical agent-of-agents orchestration, graduated autonomy, and adaptive learning pipelines.
32 agents, 249 patterns from 2,132 signals
Production-ready web applications with Next.js, React 19, Tailwind CSS v4, TypeScript, PostgreSQL, Prisma. Multi-tenant architecture, authentication (Better Auth), booking, ordering, CMS, inventory, and embedded AI agents.
30+ packages, 11 client deployments
Docker deployment, nginx reverse proxy, SSL automation (certbot), health monitoring (57 endpoints via Gatus), Telegram alerting. Vault-backed secret management with SOPS + Age encryption, 4-layer key audit, automated rotation.
57 endpoints monitored, zero-downtime deploys
Grounded AI advisors for professional services. Hybrid search (FTS5 + vector + reciprocal rank fusion), LLM query expansion, domain-aware re-ranking, source citations. Deployed for tax, hospitality, e-commerce, and customer service.
31/31 QA pass, 13,088 knowledge chunks
End-to-end brand identity: logo design, color systems, typography, design tokens, component libraries. 16 design systems built across client and internal products. Brand guide generation from PDFs and design documents. Interactive HTML brand kits.
16 design systems built, 75+ components
Brand-consistent marketing content at scale. Image generation (Gemini, Nano Banana), video generation (Kling, Veo 3.1, SkyReels), voice (VoiceBox, ElevenLabs), lip sync (Hedra, HeyGen). Full pipeline from creative brief to final render.
9,276 indexed prompts, 7 generation backends
The platforms, languages, and tools behind every product.
Solo operators hit a ceiling. There's only so much you can hold in your head, so many files you can track, so many decisions you can remember.
I built claude-PM to remove that ceiling. It's a 32-agent system where each agent has a defined role: one researches, one codes, one deploys, one tests, one reviews, one handles security. They don't step on each other. They follow rules. They have budgets.
But the real power is that the system learns. Every commit, every bug fix, every architectural decision becomes a signal. Those signals get synthesized into patterns. 2,132 signals in. 249 reusable patterns out. The 50th project benefits from every lesson of the first 49.
The Adaptive Agent Loop captures signals automatically on every commit, runs daily synthesis across all repositories, and promotes agents from supervised to autonomous operation based on track record. Cross-project patterns are detected and proposed for review.
This is how one person delivers at team scale.
Specialized Agents
Deployed Skills
Signals Captured
Patterns Learned
Documentation Files
6 of 32 specialized agents. Each has its own tools, rules, and context window.
Routes tasks to the right agent based on intent. Manages cross-agent dependencies, budget allocation, and parallel execution across 31 subordinate agents.
Discovery-first PM workflow: capture requirements, generate PRDs, break down specs into tickets, plan phases, and verify acceptance criteria. 10-phase pipeline from idea to shipped feature.
Automated test generation, acceptance criteria verification, regression detection, and pre-merge validation. Ensures every feature meets spec before it ships.
Single source of truth for all deployments, CI/CD, secrets management, and infrastructure. Enforces staging-first, pre-deploy key audit, and zero-downtime rollouts.
Deep scanning for secrets, injection vulnerabilities, and OWASP patterns. 4-layer key audit across vault, GitHub, VPS, and local environments.
SOPS + Age encrypted vault. Automated rotation, cross-project key sharing, orphan detection, and pre-deploy validation gates.
72 custom skills loaded on demand. These are the heavyweights.
Structured product requirements from ideas, research, or problem statements
Decompose specs into epics, stories, and tasks with acceptance criteria
Group tickets into executable sprints with QA gates and deploy strategy
Transform PRDs into implementation-ready technical specifications
Context-aware UI guidance with Distinctive and Functional modes
Prisma schema drift detection and data loss prevention
Mandatory pre-deploy validation with secrets audit and staging gate
Compose file validation against 30+ documented gotchas
4-layer vault audit across GitHub, VPS, MCP, and local
Comprehensive technical SEO validation with Lighthouse, structured data, and OG tags
Describe your project or challenge. I'll get back within 24 hours.