Tripp Cashel · A solar-system career tour
Curiosity.
Then, make it work.
I’m a Principal Engineer at Magnite. I build software systems, work with teams, and make tools for working with AI.
A scenic route through my career, from electrical engineering to software and AI. Scroll to travel from the Sun to Eris.
2011 → Now ↓201101 / 09
Boston University
Start with the physical world.
I studied electrical engineering. My undergraduate project involved reverse engineering an undocumented camera protocol in C — a first bridge between hardware and software.
More about this work ↗2012–201502 / 09
PW Power Systems
Engineering, in the field.
I engineered vibration monitoring and generator breaker protection for gas-turbine plants, diagnosed problems at operating sites, and led technicians through a fleet hardware upgrade.
More about this work ↗2016–202103 / 09
SpotX
From power plants to real-time software.
I joined the team moving an advertising auction from PHP to C++. One service I reverse engineered and rebuilt became limited by the server’s network interface. I also built benchmarking tools and introduced compiler caching.
More about this work ↗201704 / 09
University of Denver · M.S. Electrical Engineering
When the edge of the grid makes decisions.
My master’s thesis modeled nodal power flow and coordinated dispatch on a 33-bus network. I studied the transition from centrally managed resources to distributed participation, including how reliability changes prices—and when the coordination loop fails to settle.
Read the research ↗2020–202105 / 09
SpotX · Senior engineer & technical lead
Own the change across the system.
I led the design and implementation of an auction-pricing change spanning the auction itself, beacon generation, and logging. The work required coordinating the behavior of several parts of a live system.
More about this work ↗2021 onward06 / 09
Magnite · Staff engineer & team lead
Build the team’s ability to deliver.
I led and managed a team working on auction logic and data-science algorithms in C++. I worked with product on requirements, planned implementation with the team, and improved compiler, build, development, and QA environments.
More about this work ↗2025–202607 / 09
Magnite · Agent-platform architecture
A shared tool layer for agents.
I architected the original prototype agent platform and gateway system through discussions with my manager and teammates. Ryan implemented the AI gateway. I designed the MCP gateway so one platform could use teams’ tools for internal and customer-facing work. Some of that design continues in the newer platform.
More about this work ↗January 1, 202608 / 09
Magnite
Principal Engineer.
I became a Principal Engineer on January 1, 2026. My work now connects agent-platform architecture with the practical questions of operating it: identity, access, and how teams share tools.
More about this work ↗202609 / 09
Magnite · MCP identity & access
Carry the right identity through.
I worked with the team on OAuth integration, downstream identity propagation, human and agent calls, rate limiting, and durable audit logs. The gateway needed to preserve who was making a request, and in which account context.
More about this work ↗Alongside the day job
Still making things.
Juice makes coding-agent history searchable. Forge coordinates coding agents. AiChat is where I experiment with what an agent can make possible — including this site.
Try the AI-generated portfolio ↗About this solar-system journey
The Sun, eight planets, Pluto, then Eris. Their equatorial diameters share one scale, but viewing distance changes for each encounter. Travel spacing is compressed; the route does not represent current orbital positions. Lighting, rotation speed, and the star background are composed for the journey.
The lines illustrate a Parker interplanetary magnetic field carried by a steady solar wind, plus centered dipoles for Mercury, Earth, Jupiter, Saturn, Uranus, and Neptune. They trace the magnetic field, not particle trajectories. Near the Sun, the outer field is shown with logarithmically compressed distance.
Glow and moving highlights make the traced lines easier to follow; their brightness and speed are visual cues, not field strength or particle motion.
This is an analytical field illustration, not a full plasma simulation: it omits magnetopause currents, bow shocks, reconnection, and induced fields at Venus and Mars. No intrinsic field is assumed for Pluto or Eris.
Parker field equations ↗ · NASA body and dipole data ↗
Planet textures: Solar System Scope, CC BY 4.0 (color-adjusted maps; Eris is fictional). Pluto: NASA / JHUAPL / SwRI, New Horizons mosaic with uneven coverage.