SFTP on the go

A lightweight, dockerized sftp server with basic authentication and security

SFTP is the most direct method of sending data. As an alternative to SaS cloud hosting, local-hosting is a means of warehousing your own data. By staying lightweight with a limited Debian OS, and leveraging docker containerization, this service can run on a Raspberry Pi with dedicated SSD or HDD storage. Stay tuned for improved security features, and automated backup functionality.
  • Lightweight, Debian-based sftp server ready-made for Docker deployment.
  • Includes basic user authentication
  • Allows direct SSH and TLS connections.
As the entire build is containerized with Docker, this service can be build anywhere with a Docker or Kubernetes, including Raspberry Pis Includes template rules for IP whitelisting / rate limiting for iptables.

STATFOX - PYPI

A Python module for advanced data science procedures.

Functionality currently includes:

  • Convenient distribution resampling for robust statistical parameter estimates.
  • Kernal-aided probability density function estimation
  • Arbitrary function maximization procedures with MCMC

Next-generation Astrophysics with Artificial Intelligence

Machine learning is poised to significantly expand the capabilities of observational astronomical research. I have designed a generative artificial neural network model for use with the Southern Photometric Local Universe Survey, demonstrating an unprecedented capability to estimate the metal and carbon abundances of stellar atmospheres.

Devin Whitten

Training on photometric inputs for estimation of Absolute Carbon Abudance, using monte carlo generative training procedure. Activation functions and hidden layer size are optimized within a randomize hyperparameter space.

Interactive Network Architecture

Bayesian Carbon Modeling in CEMP Stars

Developed pipeline for determination of chemical abundances in stellar atmospheres

Devin Whitten

Bayesian spectral matching is a revolutionary method to estimate stellar parameters for among the most strongly depressed stellar atmospheres. By implementing a procedure that weighs the influence of multiple molecular and atomic features associated with the presence of CH, C2, and CaII, stellar parameter determination can appropriately mitigate the effects of carbon and metal-line blanketing. This technique promises to provide parameter estimates for countless stellar sources at low- and medium-resolution, previously inaccessable by alternative methods.

Gaussian Inflection Spline Interpolation Continuum

Automated spectral normalization

Devin Whitten

Intuitive spectral normalization procedures are crucial to automate the stellar parameter determination process, considering the overwhelming wealth of spectrscopic data being acquired by multi-fiber survey expeditions. The idenfication of strong molecular depression in the spectral energy distribution is essential to provide accurate continua, and by extension, chemical abundance determinations. I have designed such a procedure, based on 2nd order numerical differentiation of spectral convolutions. This procedure has made possible the first detection of a Group III carbon-enhanced metal-poor star in the Canes Venatici I dwarf galaxy.