We take on problems where signal processing meets shipped software: designing the algorithm, proving it works with measurements, and building the production application around it. If your problem involves extracting decisions from signals — audio, RF, or sensor data — this is our home ground.
Detection, classification, and estimation algorithms designed from your data: FFT-based spectral analysis, digital filtering, noise reduction, feature design, and statistical decision logic. Every algorithm ships with a validation report — test methodology, accuracy numbers, and known failure cases — so you know exactly what you're deploying and where its limits are.
End-to-end software-defined radio solutions, from waveform to working application. We have designed and implemented transmit and receive chains for MPSK/MQAM, OFDM, and CDMA waveforms, and signal detection algorithms for activity, frequency, and symbol-rate estimation. We integrate SDR hardware — NI USRP, ADALM-Pluto, HackRF, BladeRF, RTL-SDR — for real-time IQ acquisition and processing, build automated IQ data-collection and preprocessing pipelines for training AI signal-classification models, and optimize DSP pipelines to meet real-time constraints.
Audio authenticity and quality analysis, productized. Transcode and fake-lossless detection, batch QC tooling for large audio libraries, and audio feature extraction for machine-learning pipelines. Our free product Sunara Reference is this capability in shipped form — a five-detector evidence-fusion engine validated at 96.5% on a controlled paired dataset — and the same engineering approach extends to delivery-spec QC and catalog-scale authenticity screening.
Algorithms only create value inside software people can run. We build real-time desktop applications, web applications, and mobile apps — and we specialize in the unglamorous part: taking a working prototype or research algorithm and turning it into a fast, maintainable product with a clean interface. DPI-aware Windows desktop work in C#/.NET is a particular strength.
Engagement
What you're trying to detect, measure, or build; what data you have; what a good outcome looks like. A short email is enough to start.
Before committing to a build, we examine your data and constraints and tell you honestly what's achievable — including when the answer is "this won't work, and here's why."
Development in phases, each closing with something you can run and numbers you can check.
Documented code, a validation report, and the knowledge transfer your team needs to own it.
Fixed-scope projects and hourly consulting both work — whichever fits your problem. We contract directly (email below) or through Upwork if you prefer platform protection: Upwork profile. NDAs are no problem.