Thoughtfully designed products combining AI, privacy, and modern engineering into tools built for practical use.
We engineer production-grade local-first utilities, automated community safety suites, and serverless patent analytics.
PatentIQ is an automated screening tool designed to streamline patent transitions. By linking a serverless Cloudflare Workers backend with an Express LLM microservice, the platform parses patent documents, validates prior-art citations via SerpAPI, and delivers a rigorous 11-dimension patentability score. Features an interactive dashboard and dynamic, on-demand clientside PDF generation.
Dia Chat is a local-first conversational interface designed to establish absolute user privacy. By leveraging browser-native WebGPU acceleration, the application executes WebLLM and Transformers.js models directly within the user's environment. This enables high-performance real-time chat, voice synthesis, and session storage via Firebase without ever exposing sensitive interaction logs or identity data to external AI servers.
Dia Moderator is an automated safety layer designed to consolidate fragmented community management. By linking a unified dashboard with lightweight bot connectors for Discord, WhatsApp, and Telegram, the platform intercepts and analyzes social streams in real-time. It mitigates toxic behavior and filters automated spam using Gemini AI, while delivering instant WebSocket-powered activity metrics and community health scoring directly to administrators.
Interested in our developer libraries, helper utilities, or experimental codebases? We build everything in the open and welcome contributions to our repositories.
Three obsessive builders engineering products combining AI, privacy, and modern systems architecture.
Builds the robust core architectures, serverless frameworks, and deep ML pipelines. Prefers making complex systems scale silently in the background.
Transforms complex, heavy data structures into gorgeous, fluid frontends. Believes high-performance tech deserves a premium screen presence.
Squeezes maximum efficiency out of NLP queries, semantic indices, and custom algorithms. Finding the fastest path from raw data to deep insight.