Skip to content
View Hitheshkaranth's full-sized avatar

Block or report Hitheshkaranth

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

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

Report abuse
Hitheshkaranth/README.md

Hithesh Karanth. From silicon to intelligence, built to be verified. A command menu lists the stack: silicon, firmware, avionics, assurance, tools and intelligence.

FlyVI Technologies  ·  Noyce IDE  ·  OpenTerminalUI  ·  @HitheshKaranth


About

I lead technology at FlyVI Technologies, building high-reliability aerospace, defence, power-electronics and embedded-control systems. Outside of that I build developer tools, financial infrastructure, applied-AI products and self-hosted LLM inference, mostly in the open.

Aerospace and defence, embedded systems, developer tools, applied AI, MLOps and inference



Building now Noyce IDE: an AI-native Code-OSS workbench for safety-critical firmware



Recognition WireVoice, winner of the Sarvam Epoch Buildathon 2026: a voice-first assistant for wiring harness drawings



Open source OpenTerminalUI: a self-hosted financial terminal OpenTokenMonitor: a local-first monitor for AI coding tool usage EmbeddedDisplayStudio: ship HMI screens to embedded Linux panels



Systems work ARINC 615A Tool Suite: avionics software loading over Ethernet NETRA System Debugger: telemetry diagnostics and a live 3D digital twin NETRA Device Firmware: ESP32-C6 ultrasonic ranging and motion telemetry



Inference Three DGX Spark deployments: Ornith-1.5 at 351.7 tok/s across 16 users, Qwen3.8 Distill at 373 tok/s across 16 users, Qwen3.6 at 219 tok/s peak, all at 262K context Ornith-1.5-35B-A3B on DGX Spark Qwen3.8-35B-A3B Distill, self-quantized Qwen3.6-35B-A3B-NVFP4 production recipe



Stack Stack, from silicon to intelligence: C, STM32, ESP32, LVGL; C++23, Rust, Qt 6, Embedded Linux, VxWorks, ARINC 615A; DO-178C, MISRA, CBMC, CodeQL, MC/DC; TypeScript, React, Tauri, Code-OSS, FastAPI, PySide6; vLLM, NVFP4/FP8, speculative decoding, voice AI, agents



Star history

Star history for OpenTerminalUI, OpenTokenMonitor, EmbeddedDisplayStudio and Ornith-1.5


X FlyVI Technologies website Repositories Follow on GitHub

Open to collaboration on developer tools, financial infrastructure, voice AI, self-hosted inference, avionics and embedded systems.
The best way to start is an issue or discussion on the relevant repository.

Thanks for stopping by.

Popular repositories Loading

  1. OpenTerminalUI OpenTerminalUI Public

    OpenTerminalUI — a trading terminal UI for market data, charting, screening, backtesting, and alerts.

    TypeScript 282 59

  2. Ornith-1.5_A3B_Model_DGX_Spark_Setup Ornith-1.5_A3B_Model_DGX_Spark_Setup Public

    Ornith-1.5-35B-A3B (NVFP4 + FP8 + MTP) on NVIDIA DGX Spark (GB10)

    Python 29 3

  3. OpenTokenMonitor OpenTokenMonitor Public

    OpenTokenMonitor is a lightweight, local-first desktop widget for tracking AI CLI usage across Claude, Codex, and Gemini. It helps developers monitor activity, usage trends, and costs in real time …

    Rust 24 1

  4. EmbeddedDisplayStudio EmbeddedDisplayStudio Public

    Describe, draw or bring your own HMI screen and ship it to embedded Linux panels: a desktop Studio with AI Design and a coding agent, a Qt-free C + LVGL panel runtime, and atomic SSH deploys with a…

    Python 16 1

  5. Netra_Device_Firmware_V1 Netra_Device_Firmware_V1 Public

    ESP32-C6 firmware for NETRA ultrasonic ranging and MPU6050 motion telemetry

    HTML 3

  6. Qwen-3_8_A3B_Model_DGX_Spark_Setup Qwen-3_8_A3B_Model_DGX_Spark_Setup Public

    Quantize (NVFP4 experts + FP8 attention, data-free) and serve Qwen3.8-35B-A3B-Distill with vLLM on a single NVIDIA DGX Spark (GB10) — 262K context, 16 concurrent users, 373 tok/s aggregate.

    Python 3