I am a Principal Data Scientist and Manager at The RepTrak Company.

Before that, I was a Research Scientist at Luminoso Technologies, Inc, where the focus of my research was Neural Machine Translation (NMT).

Even before that, I was a Postdoctoral Research Associate at Brookhaven National Laboratory (BNL), where I participated in the activities of the Electronic Detector Group (EDG) in the Physics Department as an active member of three large, leading-edge, international experiments; DUNE, SBND, and ICARUS.

I am the author of the poetry book, infinite void, finite feelings

I am the creator, developer, and maintainer of

I received my B.S. degree in Physics from Minnesota State University, Mankato and a Ph.D. degree in Nuclear Physics from Iowa State University. My graduate research work focused on studying Quantum Chromodynamics (QCD) through matter under extreme conditions using the PHENIX experiment at the Relativistic Heavy Ion Collider (RHIC) at BNL. My Ph.D. research work has been awarded Research Excellence Award by Iowa State University in recognition of the outstanding research accomplishment in a graduate program, and my Ph.D. dissertation has been awarded the RHIC and AGS Thesis Award in recognization of the most outstanding thesis related to research conducted at BNL's facilities.

You can find my dissertation here.

You can find the list of my publications and citations summary on INSPIRE.

Old Machine Learning Projects

Let Jerry Seinfeld finish an incomplete sentence

  • GPT-2 is fine-tuned on dialogues of Jerry Seinfeld from 9 seasons of Seinfeld scripts using PyTorch and Huggingface's Transformers; code is here.
  • The trained model is deployed in Google Cloud Platform; code is here.

Try it here:


Particle Class Separation:

  • Semantic segmentation (pixel-wise classification) network to perform cosmic ray and beam particle separation in prototype DUNE detector.
  • The network achieved Intersection-Over-Union (IOU) of 0.94, averaged over three classes.

Arbin Arbin

Press the play button to see model's prediction improve over training time.