Back to home

Our Work

Case Studies

Every project below is in production, handling real users and real transactions. These are systems we designed, built, and shipped.

Marine Services / Custom Business Platform + SEO & Analytics

Charter Operations Platform

From phone calls to a full-stack booking system

Charter Operations Platform
Next.jsTypeScriptSupabase

The Challenge

A Cape Cod charter fishing captain was managing his entire operation through phone calls, text messages, and a basic website builder. During peak season, he was losing bookings to scheduling conflicts, spending hours on admin instead of the water, and had zero visibility into his business performance.

What We Built

We built a complete operations platform from scratch: a public-facing site with trip information, real-time availability calendar, and inquiry forms, all backed by an admin dashboard with a full booking pipeline, automated email notifications, gallery management, and business analytics. The system integrates NOAA tide data and weather APIs to provide cancellation probability predictions and optimal trip timing recommendations.

Key Deliverables

+39 API endpoints powering the entire operation
+Automated booking pipeline: inquiry to discussion to deposit to confirmed
+Tide and weather integration for trip planning
+Admin dashboard replacing all manual tracking
+Full SEO implementation with structured data
+Automated health monitoring and maintenance
Next.jsTypeScriptSupabaseNOAA APIResend

B2B Construction Materials / AI-Powered Tools + Cross-Border

Trilingual Product Intelligence

84 products, 3 languages, one AI-powered catalog

Trilingual Product Intelligence
FastAPIPythonChromaDB

The Challenge

A waterproofing materials supplier needed their B2B buyers across English, Chinese, and Spanish-speaking markets to find the right product for their specific application, substrate, and budget. Traditional catalog browsing meant buyers either called a rep or guessed wrong.

What We Built

We built a RAG-powered product intelligence system: an AI chat that answers natural-language questions about 84 waterproofing products, a 5-step guided product finder wizard (application to substrate to scale to budget to recommendation), and a filterable product catalog. The system uses hybrid search combining vector embeddings with BM25 keyword matching and reciprocal rank fusion for high-precision results across all three languages.

Key Deliverables

+Hybrid RAG search with vector + keyword fusion
+Trilingual support across UI, backend, and documents
+5-step guided product recommendation wizard
+248+ automated tests ensuring reliability
+Deployed on Railway with auto-deploy from GitHub
+Full API with structured product data
FastAPIPythonChromaDBBM25Railway

Developer Tools / Productivity / AI-Powered Tools

Document Processing Toolkit

18 pipelines, AI chat, and zero cloud dependency

Document Processing Toolkit
PythonFlaskPyPDF2

The Challenge

Document processing is fragmented: one tool for PDF compression, another for image conversion, another for merging, another for OCR. Each requires uploading sensitive files to a third-party server. Businesses dealing with confidential documents need processing power without the privacy risk.

What We Built

We built a comprehensive, privacy-first document processing suite with 18 registered pipelines, all running locally. PDF operations (compress, merge, split, rotate, watermark, protect), image operations (resize, convert, background removal via AI), and an AI chat assistant where you describe what you need in natural language and the system generates and executes deterministic processing plans. Includes a RAG document engine for querying uploaded documents with hybrid BM25 + TF-IDF retrieval.

Key Deliverables

+18 document processing pipelines
+AI chat: describe tasks, auto-generated execution plans
+RAG engine for document Q&A with hybrid retrieval
+Privacy-first: everything runs locally, zero cloud uploads
+Recipe system for chainable multi-step workflows
+Packaged as standalone .exe for end users
PythonFlaskPyPDF2OpenCVONNXSQLite

Your project could be next.

Tell us what you need built. We'll tell you how we'd approach it.