Showing 30 of 1,431 · Category: Machine Learning
Find the one-character Adam bug, build FlashAttention from the online-softmax recurrence, size a 70B model's KV cache three ways. Tested NumPy companion to the book Math for ML Engineers: From Backprop to RLHF. Free Chapter 5 PDF inside.
Finance Forecast Suite - FinPredict (real-time forecasting standalone exe) + FIN_Terminal (offline multi-market research terminal): 42-factor calibrated ensemble, multi-agent AI committee, honest accuracy audit
Top ML Model Serving (Opensource) Star if you like it!
Thehro: detects when a bank customer is paying under pressure from a digital arrest scam call, and holds only that payment. Android app + FastAPI scoring service. Team ThreatForge, RAKSHAM AI Hackathon.
A Streamlit-based AutoML application for classification and regression using Scikit-learn.
Альфа Хакатон. Умный кредитный кабинет — Backend / ML portfolio: FastAPI, PostgreSQL, Docker, reproducible risk modeling.
Full-stack Django store built on 1M+ real retail transactions: customer segmentation, repurchase prediction, market basket recommendations, Plotly dashboard and Stripe test-mode checkout
Weekly prices and stock of the most popular PC parts at US retailers: graphics cards, CPUs, SSDs, hard drives, memory and power supplies. Excel and JSON, CC BY 4.0.
A blazing-fast, JAX-accelerated Cellular Automaton engine for causal discovery and DAG structure learning.
A Python semantic cache for LLM applications with similarity-based retrieval, SQLite storage, optional HNSW vector search, and safeguards against incorrect cache hits.
Unofficial Rust port of Spotify's Basic Pitch: polyphonic audio-to-MIDI transcription with pitch bends. Pure Rust inference with tract or ONNX Runtime, bundled model, CLI and WebAssembly. Parity-tested against the Python implementation, note for note.
🏆 Winner, Accenture GenAI Mavericks 2026. Synthetic population of Spain (25,000 profiles) that predicts survey answers by region, sex and age, with a validated confidence traffic light. CIS-trained models + local LLM.
Ava: a brain-like language model in pure Rust. Spiking neurons trained by gradient descent, sparse k-winner codes, no attention, no position embeddings. Dependency-free CPU stack plus a CUDA trainer with a fused spiking kernel. Tiny today, built to scale. Source-available (PolyForm Noncommercial).
ML classifier that predicts security flaws in cryptographic protocol designs using structural features like nonce usage, mutual authentication, and key freshness checks, trained and tested on real protocols including Needham-Schroeder, WPA2, TLS, and Kerberos
HeartAI is an end-to-end, machine learning-driven Clinical Decision Support System (CDSS) designed to predict cardiovascular risk levels and assist healthcare professionals in early-stage diagnosis. By bridging advanced data mining algorithms with a modern web architecture, HeartAI delivers actionable diagnostic insights from clinical data.
Laya (open-source Jev) vs classic ML, local LLMs and GPT on 1,800 real financial complaints. Accuracy, latency, cost and calibration, with paired significance tests
Real-time driver drowsiness detector using MediaPipe, a genetic algorithm, KNN, K-Means and a NumPy autoencoder
ML regression app that predicts a mental health score from social media usage, lifestyle & academic data. FastAPI + React.
Classical versus quantum machine learning for credit default prediction: an honest benchmark at matched sample sizes on 30,000 real credit-card clients. MQFS working paper.
NLP-powered fake news detection and credibility scoring system using machine learning.
PKU-AFT quantitative finance paper wiki: 37 papers across factor mining, forecasting, portfolios, risk and execution.
Segmentación de jurisdicciones argentinas por viviendas habitadas con K-Means
AI real estate platform: home price prediction (scikit-learn), explainable property recommendations, and a RAG + Groq LLM assistant. Built with FastAPI and a web UI on the Ames housing dataset.
Offline paired evaluation audits: reveal subgroup regressions hidden by improving model averages.
Application mobile Flutter de reconnaissance de fruits par IA embarquée (TensorFlow Lite, inférence on-device hors-ligne), avec authentification Firebase et assistant vocal Google Gemini.
In this project, we’ll build a multi-agent system that collects news, weather, finance, and media data, then uses an LLM to generate a daily brief, delivered via a Streamlit app with real-world API integration.
Local-first AI document assistant featuring vector search, HNSW, KD-Tree, SQLite persistence, RAG-based retrieval, and optional Ollama integration.
Hands-on pandas practice covering DataFrame basics, data exploration, filtering, and reading/writing CSV, JSON & Excel files — part of my AI/ML learning journey.
Аренда GPU-серверов по минутам: RTX 4090 от 25 ₽/ч, RTX 5090, A100, H100. Оплата СБП и криптой (USDT, TON). ComfyUI, Stable Diffusion, Ollama, генерация картинок без цензуры — gpu.nz
Modelo de previsão de inadimplência