What I Built for
My Own Companies

The Problem

The founder was posting content at 11 PM, manually checking Amazon reviews every morning, and losing Buy Box to competitors for days before noticing. Customer comments sat unanswered for 48+ hours. Inventory ran out twice because nobody was tracking FBA stock levels. A 5-person team (content manager, social media, customer service, analyst, ops) would cost $300K+/year. He had product-market fit but was drowning in operations.

What I Built

• Multi-platform content engine: Publishes 35-40 posts per week across YouTube Shorts, Instagram Reels, Facebook Reels, and TikTok. 233 videos in rotation. Instagram Stories 3x/day. Zero manual posting.
• AI customer engagement layer: Automated comment responses on 3 platforms using Claude API. Spam filtering, brand voice consistency, contextual replies. Runs 3x daily, 365 days/year.
• Amazon intelligence system: Real-time review monitoring (700+ reviews, 4.6 star average), Buy Box tracking 4x/day with competitor hijack detection, daily automated P&L delivered via Telegram at 9 AM.
• Competitor price monitoring: Detects price changes, new sellers, and Buy Box losses. Alerts only when action is needed.
• Inventory forecasting pipeline: Tracks FBA stock levels, in-transit units, reserved quantity. Flags reorder points before stockouts hit.
• Unified command center: Telegram bot consolidating all alerts, reports, and controls into one interface. The founder runs the entire business from his phone.

Results

Timeline

Tech Stack

The Problem

Product was sitting in a Cleveland warehouse collecting dust. No leads, no CRM, no website, no online presence. The founder was cold-calling from a spreadsheet and getting nowhere. Hiring a sales team (2-3 reps + manager) would cost $250K+/year and take 3-6 months to ramp. Meanwhile, warehouse rent was burning $3K/month on unsold inventory.

What I Built

• AI-powered lead generation engine: Built automated scraping system targeting 12 buyer segments (stone yards, contractors, architects, pool builders, landscape architects). Collected 29,207 qualified leads with business name, phone, email, address, rating, and reviews. 11,534 leads enriched with verified email addresses.
• AI voice agent for outbound qualification: Deployed conversational AI agent ("Alex") on ElevenLabs with custom voice profile, 40+ call scenarios, and objection handling scripts. Handles cold outreach calls, qualifies leads, and warm-transfers serious prospects. Sounds human, not robotic.
• 25-page SEO-optimized website: Built complete web presence: product pages, 4 application pages (driveways, patios, pool decks, commercial), 5 geo-targeted state pages, 11 SEO blog articles with full schema markup. All pages indexed and ranking.
• Automated social media presence: Facebook autoposting system running Mon-Fri with A/B tested hashtag sets. AI-powered autoresponder handles page comments 2x/day using Claude.
• B2B email outreach system: 4 audience-segmented email campaigns with personalized templates, merge fields, and automated follow-up sequences. Connected to Instantly.ai for delivery.
• Autonomous sales operations: AI handles market research, competitor analysis, email composition, and sales forecasting. The founder focuses on closing deals, not chasing leads.

Results

Timeline

Tech Stack

The Problem

The founder was paying $2/lead from ZoomInfo for generic contact lists with 30% bounce rates. At 146K targets, that's $292K only for names and emails. Worse, the lists had no buying intent data. He was blasting the same message to everyone and getting ignored. Meanwhile, competitors with bigger sales teams were locking up NYC contractors one by one.

What I Built

• 8-source lead intelligence system: Integrated NYC Open Data APIs (DCA licensed contractors, DOB building permits, HPD paint violations, HPD managing agents, job filings, license applications) plus Google Places and Yelp Fusion. 146,000+ leads collected, deduplicated, and enriched with 27 data fields each.
• Pain-based segmentation engine: Leads scored and segmented not just by industry but by buying intent. HPD paint violation data identifies property managers who must repaint (compliance deadline). DOB permits identify contractors mid-project. This is intelligence, not a contact list.
• Full autonomous sales operations: AI orchestrator routes tasks to specialized agents: lead finding, email personalization, bounce prevention, market research, competitor tracking, and forecasting. The system runs the entire sales pipeline without manual intervention.
• 20-template email outreach system: 4 emails per segment, 5 segments. Personalized by borough, business type, and pain point. Subject lines optimized for open rate. Connected to Instantly.ai for automated delivery and tracking.
• Sales simulation system: Built 96 realistic prospect personas (12 companies x 8 variations) to test email sequences before launch. Forecast conversion rates, identify objection patterns, and optimize messaging pre-send.
• Single reporting dashboard: Automated daily digests to Notion: lead counts by source, enrichment progress, campaign metrics, pipeline status. Full visibility without manual reporting.

Results

Timeline

Tech Stack