Skip to content
Contact
Contact

Where we go deep

We organize our experience by type of problem, not by industry or technology. Techniques change: signal processing, machine learning, deep learning, language models. What remains is the ability to go from problem to production.

Sensing & Signal Intelligence

Turning raw signals into reliable information, at scale and under noise.

Typical problems:

identifying what is in a signal when it is long, noisy or arrives in very large volumes; recognizing patterns in audio, voice, biomedical signals, sensors and time series; detecting where events start and end; telling apart nearly identical variations.

Proof:

  • ECAD, Brazil's central office for music copyright collection. Since 2008, identifying music across about 4,000 radio stations and 400 TV channels, and in live performances. Today: close to 100% accuracy on radio, with a fully automatic process, and 98% on TV. An evolution from classical signal processing to orchestrated neural networks, without ever stopping operations. Read the case

Applied AI & Knowledge Engineering

Applying AI to specialized knowledge, knowing where it adds value and where it adds risk.

Typical problems:

bringing the knowledge held by experts, documents and systems into day-to-day decisions; adapting AI models to a domain, language or repertoire they were not trained for; deciding where generative AI helps and where the calculation must be deterministic and auditable.

Proof:

  • Radix · supported by Finep. A partnership in progress in a 36-month research and development program for a project management copilot. The feasibility research changed the architecture before construction began. Read the case
  • ECAD. Identifying live music, including covers and songs in other languages, by adapting neural networks to the Brazilian repertoire. Read the case

Beyond these areas

These capabilities are a starting point, not a fence. If your problem is complex, relevant, has no off-the-shelf solution and needs research and engineering to be solved, we want to hear about it.

Tell us about the problem you haven't been able to solve yet