MiniGPT-M4
A compact language-model study exposing Transformer internals through a custom BPE tokenizer, PyTorch components and local Apple Silicon training experiments.
PROJECT INDEX / EN
Explore research, prototypes and early work with their real development context. 22 repositories and related independent work, each with its own scope and maturity.
A compact language-model study exposing Transformer internals through a custom BPE tokenizer, PyTorch components and local Apple Silicon training experiments.
Private local-first finance application combining double-entry accounting, reviewable document imports, payment planning and encrypted backup support.
A Streamlit prototype combining Turkish tender PDF extraction, AI-assisted risk review, retrieval-based questions, comparison and reporting components.
A local-network prototype turning an iPhone into a Mac trackpad, keyboard and system-control interface through a shared Swift protocol and WSS.
A modular prototype organizing authorized open-source investigations through plugin scans, persistent jobs, an API, live progress and optional worker processes.
Private document-processing prototype exploring local PDF-to-table extraction through batch processing and a desktop interface, with explicit privacy boundaries.
Private research toolkit exploring EEG, signal features and classical machine learning through desktop, offline and local web components.
Private logistic-regression research experiment exploring binary classification, ROC visualization and explicit input validation for tabular data.
A competition archive combining time-window customer features, boosting models, ensemble blending and Optuna experiments for customer-behavior modeling.
A research collection exploring noisy Turkish-address normalization and task-specific label prediction through text methods, vector retrieval and geographic approaches.
A local-first agent core separating LLM proposals from deterministic risk policy, approval state and tool execution, with optional Apple integrations.
An alpha assistant combining Ollama and optional MLX models with rule-based routing, memory, tool experiments, voice and a React/Three.js interface.
Private SwiftUI iOS prototype exploring profiles, matching and communication around skills people can teach and want to learn.
A desktop document-preparation tool arranging front/back badge PDFs into A4 Word documents with previews, ordering and configurable printing layout settings.
A desktop utility presenting CPU, memory, GPU, disk, network and operating-system information in tabs, with JSON and HTML report export.
Private enterprise application work focused on user workflows, controlled integration boundaries and offline software validation within an established project scope.
An anonymized process-analysis study examining the relationship between manufacturing execution workflows and autonomous mobile robot operations.
Experimental computer-control prototype using MediaPipe/TFLite hand tracking, with the left hand controlling the cursor and the right hand handling clicks and scrolling.
Trendyol Hackathon work represented among competition experiences, without extending the supplied technical scope, team role or outcome details.
A source-cataloged archive of 33 independent notebooks exploring machine-learning ideas across image, text, audio and tabular data tasks.
An MVC and Razor interface prototype showing healthcare landing pages, doctor-dashboard navigation and registration/login layouts, with incomplete server-side functionality.
Private archive preserving earlier Kotlin/Android and Flutter implementations of RiseMate’s skill-exchange idea separately from the current Swift client.
An early computer-vision prototype with two independent scripts exploring webcam motion or external YOLO hand detection for triggering media keyboard shortcuts.
A learning archive preserving original ZIP snapshots of seven Python exercises covering calculations and data, alongside twenty AutoCAD DWG drawings.
A historical source archive preserving original Android, Java and Python project boundaries across interface, algorithm and automation exercises from early learning.
No projects match these filters.