Skip to content
View calebtt's full-sized avatar
7 followers · 14 following
  • Mount Juliet, TN
  • 07:17 (UTC -06:00)

Achievements

Achievement: QuickdrawAchievement: Pull SharkAchievement: YOLO

Achievements

Achievement: QuickdrawAchievement: Pull SharkAchievement: YOLO

Block or report calebtt

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
calebtt/README.md

Hi there! I'm Caleb

🌱 Currently diving into Multi-paradigm Design in C++

📫 Best way to reach me: Email preferred. Cell available by request for faster back-and-forth. NashDev Slack for Nashville locals.

📍 Nashville, TN – Always up for coffee chats on .NET, C++, or building AI agents.

I'm a software engineer passionate about low-level systems, AI-driven tools, and cross-platform libraries. Lately, I've been building C#/.NET components for voice processing, speaker diarization, and intelligent agents, alongside some model fine-tuning. Check out my recent work below!

Spotlight Project: MinimalTextClassifier

I'm excited to highlight my latest model fine-tuning work: MinimalTextClassifier!

This is a minimal .NET ONNX-based binary text classifier with Python fine-tuning support, powered by the Deberta transformer model. It's designed for efficient text classification tasks, bridging Python for model training/fine-tuning and .NET for lightweight, cross-platform inference via ONNX Runtime.

Key Features:

  • Fine-Tuning Support: Use Python scripts to fine-tune Deberta on your custom datasets for binary classification (e.g., sentiment analysis, spam detection).
  • ONNX Integration: Export models to ONNX for seamless deployment in .NET environments—fast, portable, and no heavy dependencies.
  • Minimalist Design: Focuses on simplicity and performance, making it easy to integrate into apps or pipelines without bloat.
  • Cross-Platform: Runs on Windows, Linux, macOS via .NET.
  • Example Use Cases: Custom text classifiers for smart home devices (Alina, what time is it?), content moderation, or data processing tools.

If you're into AI model optimization or need a quick way to deploy fine-tuned transformers in production, check it out! Contributions welcome—let's collaborate on expanding it to multi-class or more advanced scenarios.

Other Featured Projects

Voice AI & Telephony Tools (Recent C# Focus)

These are my latest C# repos, emphasizing minimal, efficient libraries for speech detection, diarization, and agentic workflows. They're battle-tested in noisy environments and utilize AI models like Silero VAD, ecapa-tdnn, whisper, and phi3.

  • MinimalSileroVad
    Cross-platform .NET library for VAD (Voice Activity Detection) and speech segmentation using the Silero VAD model. Replaces WebRtcVadSharp—faster and more accurate for real-time apps.

  • MinimalVoiceAgent
    A lightweight voice agent framework with tool calling, local TTS/STT, and Grok API integration. Build conversational AI without the bloat—ideal for prototypes or local setups.

  • SipBotOpen (Highlighted Project)
    Open-source telephone voice assistant with tool calling and PSTN connectivity. Dial it live at 1-615-899-3280! Uses SIP client in C# .NET for VoIP PBX magic. Currently testing VAD, AEC, and noise suppression.

  • MinimalDiarization
    .NET library for AI-powered speaker diarization via ONNX runtime. Includes experimental intent preprocessing for command discrimination in noisy, multi-speaker scenarios—great for voice agents.

Other Highlights

  • XMapLib
    Low-level C++ library for mapping Xbox controller inputs to keyboard/mouse. Closer to the metal for gaming/input hacks.

  • StreamToActionTranslator
    C++23 CMake project for translating input streams to actionable function calls. Experimental fun with streams and actions.

More code, tools, and experiments coming soon. Contributions welcome—fork away!

Pinned Loading

  1. MinimalTextClassifier MinimalTextClassifier Public

    Minimal .NET ONNX-based binary text classifier with Python fine-tuning support. Uses transformer model Deberta.

    C#

  2. MinimalVoiceAgent MinimalVoiceAgent Public

    Minimal voice agent with tool calling, local TTS, local STT, and Grok digital intelligence via API.

    C#

  3. MinimalSileroVad MinimalSileroVad Public

    Cross-platform library for .NET VAD speech detection and segmentation based on the Silero VAD model.

    C#

  4. XMapLib XMapLib Public

    A closer to the metal library for Xbox controller to keyboard and mouse input.

    C++ 6 2

  5. StreamToActionTranslator StreamToActionTranslator Public

    CMake project, C++23, input stream to actions (function calls)

    C++

  6. MinimalDiarization MinimalDiarization Public

    C# .NET Library for speaker diarization via AI model (ONNX runtime) with experimental intent preprocessing for voice agent command discrimination in noisy environments.

    C#