evara://ai

Meet Evara — the AI that knows my work.

Not a toy chat box on this page. Evara is a prompt, primed with who I am and what I've built — open her in the AI you already trust and get answers grounded in this site. Or just read the straight answers below.

This site primes an AI An isometric diagram: a stack of pages from this site feeds a prompt into a tall AI core, which answers as Evara.
This site and llms.txt prime the prompt — your AI answers as Evara.
evara --ask "tell me about krushna"

Opens a primed prompt about my work, projects, and background — grounded in this site.

Open in

Straight answers

What does Krushna work on?

I'm an AI/ML engineer doing my M.Tech in Machine Learning at Dhirubhai Ambani University. My work runs close to the metal — LLM post-training and evaluation, GPU/CUDA programming, computer vision, and geospatial ML — and most of it ends up open-source.

What are his standout projects?

HuskML, a visual neural-network designer that exports real Keras/PyTorch code. Adrishyam, a computer-vision package for haze removal, published on PyPI. And MARGA, a GPU-accelerated bidirectional/heuristic shortest-path search on real-world graphs written in CUDA (paper under review). Exodus, an AI meeting monitor, won a hackathon.

What's his background and experience?

Recently: Frontier AI Engineer at Parsewave (LLM post-training and Terminal-Bench evals), ML Research Intern at CHARUSAT (writing CUDA kernels for graph algorithms), and ML Research Engineer at BISAG-N (Govt. of India), building flood-risk models and QGIS plugins from Sentinel-2 satellite data. I also founded HuskML.

Is Krushna available for hire or collaboration?

Yes — I'm open to interesting ML and research work, collaborations, and conversations, especially anything close to the metal (CUDA, systems, post-training) or open-source. The fastest way in is to book an intro call below, or reach me on LinkedIn.

What's his engineering philosophy?

I like knowing how things work all the way down — writing the kernel, reading the framework source, building the small sharp tool instead of importing the big one. I'd rather understand a system than wrap it.

How do I get in touch?

GitHub (github.com/Krushna-007) and LinkedIn are the best channels, or book a 15-minute intro call right here on this page.

Rather talk to the human? 👋

Book an intro call → Signature of Krushna Parmar