Projects That Moved the Needle

Every project here solved a real operational problem. I care about measurable outcomes โ€” faster workflows, more leads, and time saved.

AIAutomation

Invoice Processing Automation

Client type: Accounting Firm

Problem

The firm was spending 15+ hours per week manually extracting data from vendor invoices and entering it into their accounting software โ€” a tedious, error-prone process that delayed month-end closes.

Solution

Built an automated pipeline that ingests PDF invoices via email, uses AI-powered OCR to extract structured data (vendor, amount, date, line items), validates it, and pushes directly into QuickBooks via API โ€” with a review dashboard for edge cases.

Impact

90% reduction in manual data entry. Staff reclaimed 13.5 hours per week, redirected to higher-value advisory work. Invoice processing errors dropped to near zero.

project-preview.dev
Invoice Processing Automation

Key Result

90% less manual work

Tech Stack

PythonFastAPIOpenAI GPT-4oPostgreSQLQuickBooks APINext.js
NDA โ€” demo on request
WebSEO

EMSO โ€” Dental Clinic Website

Client type: Dental Practice

Problem

A local dental clinic was getting only 2โ€“3 new patient inquiries per week from their 2012-era website. No mobile optimization, slow load times (8+ seconds), and no clear calls to action meant most visitors bounced without contacting.

Solution

Rebuilt the entire website using Next.js for performance, designed treatment-specific landing pages optimized for local SEO keywords, added multilingual support (i18n) for broader patient reach, a live booking widget connected to their calendar, and set up Google Analytics 4 with conversion tracking.

Impact

Consultation bookings tripled from 2โ€“3 to 8โ€“10 per week within 60 days. Page load time dropped from 8s to under 1s. Site now ranks locally for 4 core treatment keywords.

project-preview.dev
EMSO โ€” Dental Clinic Website

Key Result

3x more consultation bookings

Tech Stack

Next.jsTypeScriptTailwind CSSnext-intlVercel
AI

SupportBot โ€” AI Customer Service Agent

Client type: E-commerce Store

Problem

An online store with 200+ daily orders was drowning in repetitive support emails โ€” order status, return policies, sizing questions โ€” consuming 4+ hours of the owner per day.

Solution

Built a context-aware AI chatbot trained on their FAQ, product catalog, and policies. Integrated with Shopify for real-time order lookups. Escalates to a human inbox when confidence is low. Deployed as a website widget.

Impact

80% of customer queries resolved automatically, with sub-30-second response times 24/7. Support ticket volume dropped 70%, saving the owner 3+ hours daily.

project-preview.dev
SupportBot โ€” AI Customer Service Agent

Key Result

80% queries handled automatically

Tech Stack

Next.jsOpenAI APIShopify APIPrismaPostgreSQLVercel
NDA โ€” demo on request
WebLocal Business

Comfort Air โ€” HVAC Service Website

Client type: HVAC Contractor, Austin TX

Problem

A local HVAC company ran entirely on referrals. No website, no booking system, and no way to capture leads outside business hours โ€” which meant emergency calls went to competitors who showed up on Google first.

Solution

Built a full-stack service site with a homepage, 6 SEO-targeted service pages (AC repair, heating, emergency, installation, maintenance, air quality), a server-validated booking form that emails the business instantly via Resend, and a Notion lead log for CRM visibility. JSON-LD structured data, sitemap, and breadcrumbs included for local SEO. Business details live in a single config file so the owner can update phone, hours, or service areas without touching code.

Impact

A professional web presence live on day one, with every form submission triggering an immediate email notification and Notion lead entry. The architecture is designed to hand off cleanly โ€” the client controls their content through Notion, not their developer.

project-preview.dev
Comfort Air โ€” HVAC Service Website

Key Result

Booking pipeline with automated lead delivery live

Tech Stack

Next.jsTypeScriptTailwind CSSResendNotion APIZodVercel
AutomationTool

EmailClean โ€” Bulk Email Verification System

Client type: Marketing & Outbound Teams

Problem

Raw email lists are noisy โ€” full of invalid, disposable, and catch-all addresses that hurt sender reputation, inflate bounce rates, and waste campaign spend. Manual spot-checking doesn't scale past a few hundred addresses.

Solution

Built an 8-layer email verification pipeline: syntax validation, disposable domain detection, role-based address filtering, domain typo correction, MX/DNS lookup, SPF/DMARC/DKIM checks, catch-all probing, and SMTP mailbox verification. Emails are queued through BullMQ + Redis for async processing. CSV-in, CSV-out โ€” upload a list, get clean results.

Impact

Lists of 10k+ emails cleaned in under a minute. Bounce rate risk eliminated before any campaign is sent. Operations teams get two clean CSVs โ€” good and bad โ€” ready for immediate use.

project-preview.dev
EmailClean โ€” Bulk Email Verification System

Key Result

8-layer verification pipeline

Tech Stack

TypeScriptFastifyBullMQRedisPostgreSQLDocker
AutomationTool

BulkMailer โ€” Email Campaign Sender

Client type: Newsletter Operators

Problem

Small teams running newsletters or outbound campaigns needed a self-hosted bulk email sender without the monthly overhead of full marketing platforms โ€” and the ability to send to 500k+ recipients reliably.

Solution

Built a Next.js + SendGrid campaign tool with drag-and-drop CSV upload, real-time sending progress via polling, batch processing with 1.5s rate-limit delays, and Prisma + SQLite persistence. Handles 500k emails in ~25 minutes with cursor-paginated batches of 500.

Impact

Campaign management simplified to upload-and-send. No monthly SaaS fees beyond SendGrid usage. Deployed to Vercel for small lists or Railway/Render for large-scale sends.

project-preview.dev
BulkMailer โ€” Email Campaign Sender

Key Result

500k emails in ~25 min

Tech Stack

Next.jsTypeScriptSendGrid APIPrismaSQLiteVercel
DashboardAutomation

DataPulse โ€” E-commerce Analytics Dashboard

Client type: Online Retailer

Problem

The operations team was making ad spend and inventory decisions based on daily CSV exports pasted into spreadsheets โ€” causing a 24-hour lag in decision-making.

Solution

Built a real-time analytics dashboard pulling data from Shopify, Google Ads, and Meta Ads through a Node.js data pipeline. Interactive charts for revenue, ROAS, inventory velocity, and customer LTV.

Impact

Real-time visibility across all channels. Ad spend efficiency improved 30% in the first month due to faster optimization decisions.

project-preview.dev
DataPulse โ€” E-commerce Analytics Dashboard

Key Result

30% better ad spend efficiency

Tech Stack

ReactNode.jsRechartsPostgreSQLPrismaShopify API
NDA โ€” demo on request

Have a project in mind?

These projects started with a 30-minute call. Let's talk about yours.

Book a Free Consultation