# Contact

The most direct contact route is email: **shekkizh at usc dot edu**. A short message with the relevant paper, repository, experiment, or question is easier to answer than a general introduction. For a synchronous conversation, use the [scheduling link](https://calendar.app.google/VXEYajKQhJbhwspZ6) on the home page.

## Research questions

For questions about a paper, include its title and the specific claim, equation, figure, or experimental result you want to discuss. Sarath's current topics include language-model training and post-training, reasoning and alignment, neural-representation geometry, agent-to-agent interaction, voice AI, graph signal processing, and non-negative kernel graphs. The [publications index](/publications) is the canonical list of papers and links to available manuscripts.

## Tools and website

For a problem with one of the browser tools, include the tool name, browser version, input format, expected result, and the smallest example that reproduces the issue. Do not send private source text merely to demonstrate a sanitizer or data-viewer problem; a synthetic example is usually sufficient. For a broken page or incorrect biographical detail, include the page URL and the text that should be checked.

## Other profiles

Code and public projects are available on [GitHub](https://github.com/shekkizh). Publication metadata is also available on [Google Scholar](https://scholar.google.com/citations?user=dWd8dQQAAAAJ&hl=en). Professional history appears on [LinkedIn](https://www.linkedin.com/in/shekkizh), and short public updates appear on [X](https://twitter.com/shekkizh). These profiles are useful for verification, but email is the preferred route for a direct question.

Canonical URL: https://shekkizh.com/contact