Physics · Cornell University · Class of 2028

Adam
Zachry

Descend

~600km · Magnetosphere

Physical Data
=
Useful Signals

Hello, I'm a physics student at Cornell interested in turning data from physical systems into useful information for decision-making. My work sits across space weather, meteorological forecasting and commodity markets. I've worked on radar signatures in the ionosphere, weather ensembles for power markets as well as soil-moisture signals for agricultural forecasting. So yes, basically a glorified data scientist.

~250km · Satellites

Signals, Space
and Satellites.

GNSS TEC Keogram Pipeline

Active · Hysell Lab

End-to-end pipeline studying Medium-Scale Travelling Ionospheric Disturbances. Downloads RINEX files from NOAA CORS S3, computes ionospheric pierce points at 250km, applies degree-10 polynomial detrending.

  • Phase velocities ~120–170 m/s · periods 30–50 min · wavelengths 300–450 km
  • Comparing TEC keograms against Millstone Hill & Alpena ionosonde foEs/fbEs
  • Investigating E-F layer coupling mechanisms
  • Contributed to CEDAR conference
PythonGNSSSignal ProcessingNumerical Analysis

~10km · Troposphere

Atmospheric Signals
drive Commodity Markets

PJM Weather Ensemble Model

Active

Predicting next-day PJM electricity price volatility using GEFS 31-member ensemble spread. When meteorologists disagree about tomorrow's temperature, power traders should price that uncertainty.

31

Ensemble Members

7d

Forecast Evolution Signal

HAC

Robust Errors

  • Shannon entropy of ensemble spread as volatility feature
  • Gas-power coupling interaction term
  • Quantile regression for tail risk · walk-forward validation
PythonGEFS APIOLS / Logistic RegressionHAC ErrorsWalk-forward Validation

Soil Moisture Analytics

CU GeoData

Ground sensor network analysis on soil moisture data as a Data Team member at CU GeoData.

PythonGround Sensor NetworksTime Series Analysis

Applied ML Soybean Forecasting

CU GeoData

LSTM model for soybean yield prediction, bringing sequence modeling to agricultural remote sensing.

PythonSoybeansLSTM

~0km · Ground Level

Malaysian Electricity Market Model

In Progress

Collaborating with a PhD student on real-time LMP modeling for the Malaysian electricity market. TBD.

JuliaReal-time LMPEnergy Market Modeling

Get in touch.

Open to energy trading roles, research collaborations, and conversations about chaotic systems in any form.

Emailmb2926@cornell.eduLinkedInlinkedin.com/in/adam-zachryGitHubgithub.com/AdamZachry

Built with Next.js · Three.js · Cornell, NY