DE HIMANSHU SURI·MK97FK · 144.390 MHz · APRS

Writing about the
invisible spectrum

Deep learning, SDR, and signal processing — from raw IQ samples to trained models. Field notes from the shack.

RST599
MODEFT8
PWR100W
GRIDMK97FK
// log entries
CONFIRMING QSO
QSL · BUREAUCONFIRMING QSO

Building a Local AI Tutor for Ham Radio Exams with RAG

How I built a fully local, free AI-powered study and exam tool for the ASOC examination using Ollama, ChromaDB, FastAPI, and Next.js — with a RAG pipeline grounded in the actual NIAR textbook.

DATE
2026-03-26
RST
599
MODE
RAG
ragollamafastapinextjspythonllmchromadb
TNX · 73DE HIMANSHU SURI
QSL · BUREAUCONFIRMING QSO

Listening to the Invisible: Classifying Radio Signals with a CNN

How I captured raw IQ samples from the RF spectrum, transformed them into spectrograms, and trained a convolutional neural network to distinguish APRS, FSK, SSTV, and noise.

DATE
2026-02-28
RST
599
MODE
DEEP-LEARNING
deep-learningpytorchsdrsignal-processingpython
TNX · 73DE HIMANSHU SURI
QSL · BUREAUCONFIRMING QSO

Capturing RF Signals with RTL-SDR and Python

A guide to capturing real-world RF signals from CubeSatSim using Python and RTL-SDR dongle

DATE
2026-02-22
RST
599
MODE
SDR
SDRPythonRFSignal ProcessingCubeSat
TNX · 73DE HIMANSHU SURI