Voice & Multimodal AI

A surprisingly mature full-stack project that combines fallback templates, multi-provider image generation, and model orchestration into one resilient storybook pipeline.

A Node.js orchestrator, a Flask inference engine, and structured prediction records combine into a surprisingly disciplined blueprint for accessible medical imaging.

A close look at the React Native, Expo, and Supabase stack behind a nutrition app that turns camera input into structured macro data, then gives users the final word.

How a schema-driven plugin system, GPU-first execution, and always-dirty nodes turn Nodos into a live media engine instead of a static dependency graph.

A push-to-talk companion that makes OpenClaw feel portable without opening a public front door.

openclaw-textclaw strips Apple messaging down to a relay problem, then solves it with a lean Node.js adapter, a persistent WebSocket, and just enough state to keep replies natural.

A public webhook, a private WebSocket, and ffmpeg in the middle. This repo shows that chatbot UX is really a delivery problem dressed up as AI.

How an ambitious voice-controlled portfolio hit the limits of the generative era and retreated to static code.

A deep dive into an AI exercise system that keeps pose tracking local, turns motion into finite-state logic, and uses simple math to do surprisingly disciplined coaching.

A Python vision pipeline that turns finger-spelling into a clean skeletal signal, then pushes it through a recognition stack built for real-world clutter, not lab-perfect backgrounds.

A close look at the Polyglot stack, the Mediasoup recording pipeline, and the workflow logic that turns a video call into a durable hiring system.

A deep dive into a React and Node platform where appointments, records, and AI revolve around one idea: healthcare only works when state changes are handled carefully.

A full-stack interview coach that chains fallback models, normalizes messy AI output, and scores both technical answers and STAR-style behavioral responses.

A tiny TypeScript bridge that translates browser drawing commands into portable ink traces for handwriting, signatures, and sketching workflows.

A close look at the resume-to-question pipeline, the browser-native voice loop, and the small state machine that turns mock interviews into a product.

A lean React SPA turns interview prep into a live simulation, using Whisper-powered transcription, sequential evaluation, and a deliberately simple architecture to make practice feel immediate.

A no-backend edtech prototype that uses speech recognition, local storage, PDF parsing, and phrase-first translation to help multilingual students keep up without sending data to the cloud.

A lightweight Flask and OpenCV system does more than spot inattentiveness. It logs behavior, captures proof, and turns a simple face detector into a classroom audit trail.

A React, Node, and MongoDB stack that turns resumes into structured interviews, simulates a talking interviewer with native browser APIs, and packages the result as a scored PDF report.

Gemini drafts the itinerary, OpenStreetMap grounds it in reality, and the app lets you regenerate one day or one activity without blowing up the whole trip.

A real-time aviation dashboard that turns noisy ADS-B state vectors into a calm, map-first experience by caching, normalizing, and shielding the UI from upstream failure.

A real-time Java backend that uses WebSockets for chat, WebRTC signaling for calls, and Gemini for conversation summaries, all wrapped in JWT security and rate limits.

A privacy-first wellness app that encrypts journal entries, routes AI through Supabase Edge Functions, and turns emotional check-ins into targeted therapeutic actions.

A full-stack farm management system that pairs image-based disease diagnosis with plain-spoken irrigation advice, built for farmers who may have a phone number, a leaf photo, and not much else.

A deep dive into a full-stack collaboration app that wires together Socket.io, PeerJS, Judge0, and Gemini to turn interview prep into a live, shared workflow.

An AI-powered Khata system for Kirana shops that combines Hinglish transcription, fuzzy customer matching, and rule-based risk scoring so informal credit can be tracked without forcing shopkeepers into software-shaped behavior.

A close look at the browser-native loop behind PromptHire, where Gemini generates the questions, speech APIs run the interview, and a feedback table turns every answer into a reviewable lesson.

A technical look at a prototype that splits evaluation across browser-side emotion tracking, cloud LLM scoring, and a hard three-strikes proctoring system.

How a small Python app turns a vision LLM into a leaf triage pipeline, using a strict JSON contract, an invalid-image guardrail, and a fast API wrapper around a surprisingly practical workflow.

A deep look at how EasyBudgetAI turns chat and voice into structured expenses, matches fuzzy Indian names, and uses Redis to keep AI costs, abuse, and security under control.

A Streamlit app that preserves document structure, scripts a conversational episode, and pushes the result through fast LLM summarization and neural text-to-speech.

A local React viewer that reconstructs turns, reveals reasoning on demand, and turns Gemini JSON exports into a navigable, citation-aware transcript.

A multi-service entrance monitor that goes beyond bounding boxes, using line-crossing state, snapshot evidence, and delayed notifications to convert vision into actionable alerts.

A Python pipeline for Yukkuri commentary that turns script text into rendered MP4s only after the voice exists, the duration is known, and the timeline can be locked to reality.

A Rust-and-Swift system that pairs iPhone, Mac, and LLM backends directly over Iroh, keeps the connection alive through flaky networks, and treats the phone as more than a chat window.

A Flutter prototype that predicts where you can stop, redraws the map around your position, and hands off to navigation only when the decision is made.

A tiny Celo MiniApp repo that hints at a bigger shift: mobile-native distribution, onchain identity, and personality as a product surface.

A foreground service, a broadcast receiver, and Android’s audio effects API combine to add binaural virtualization and reverb directly onto playback sessions.

A Windows desktop app in Python that keeps a live audio buffer, then stitches the past onto the present the moment you decide to save.

Instead of owning playback, this Kotlin app latches onto audio sessions, keeps a foreground service alive, and slips Virtualizer and Reverb into the signal path.

How a minimalist Python bridge uses the Gradio registry pattern to turn complex streaming text-to-speech APIs into a single composable UI block.

How a local-first daemon turned everyday messaging apps into an action-oriented AI control plane.

How a minimalist TypeScript monorepo turns the world's largest video library into a high-speed scratchpad for rappers.

How evinjohnn/era2 uses sub-second RAG and state-driven handoffs to bridge the gap between AI assistance and luxury retail.

How a minimalist Kotlin implementation decoupled the "Voice-First" monolith and brought backend discipline to the Alexa Skills Kit.

How a minimalist Chrome extension uses multimodal LLMs to turn static web text into a performance with a thick Kiwi accent.

By bridging the gap between LLM reasoning and the global telephone network, Wayfarer transforms the travel assistant from a search engine into a functional concierge.

A look inside the multimodal pipeline turning raw video frames into a real-time anxiety heatmap.