About

Eptalights Research works on one idea: Illuminating code. Connecting knowledge. Revealing wisdom. In practice that means we study how software and data actually behave, then build the tools and run the research that make them understandable, verifiable and safer to depend on.

We sit where machine learning meets systems. The work is technical and specific: program analysis, compilers, runtime protection, model fine-tuning and data engineering, done with the aim of producing results someone else can rerun and check.

What we do

Our work comes in two forms.

Products

We turn research into tools people can use. EptaKode uses AI to discover bugs and to carry out variant and dataflow analysis across codebases. EptaData brings the same rigor to data intelligence. Each has its own site, and the full list is on the products page.

Research Labs

For problems with no off-the-shelf answer, Research Labs takes on engagements in AI research and fine-tuning, code analysis and review, compiler tooling, runtime protections, data engineering and deep-tech development. We start with the evaluation, work in short cycles, and hand over tooling your team can run and own.

How we work

Measure first. Every project begins with a way to tell whether it worked: ground truth, a baseline and a number to beat.

Show the evidence. A claim comes with the script that produced it, and a result that does not hold counts as a finding, not an embarrassment.

Be sound where it matters. We use models and heuristics to prioritize, and proofs, types and exhaustive checks where a wrong answer is expensive.

Write it down. The notes are where we record what we tried, what it cost and what we would do differently.

Work with us

The way to reach us is to book a call. Describe the problem and we will tell you plainly whether it is a fit. The site also has an RSS feed if you would rather have the notes come to you.