Rithwik’s Portfolio

<aside> ☎️

</aside>

<aside> 🗺️

Career Interests

How can AI and Policy address “life or death” problems, or other injustices?
- (i.e. science, human health, disasters, geopolitics, human rights, democracy)
Human-machine interaction, hands-on building, aesthetics, brand identity, hardware

Having a sense of community, lots of genuine connection, tight-knit teams, intellectual conversations with minimal jargon
I thrive in crises </aside>

<aside> 🏗️

UC Berkeley (Computer Science, Cognitive Science). Spent two gap years on:
AI Research: Joshua Bloom’s lab working on AI+Science.
- First-author publication at ICLR on generative modeling of scientific imagery.
- Writing software for a CubeSat constellation launching in 2026.
- PhD-level intern training 3D reconstruction models for robots at Zipline
Policy Work:
- Founding Member of a nonprofit startup, the Paragon Fellowship, raised $150K in grants to do pro-bono tech policy work for local and state governments.
- Commissioner for the City of Berkeley.
- Open-source intelligence analyst for the Berkeley Human Rights Investigations Lab.
My Resume (click arrow to expand) </aside>

<aside> 🧠

Methodologies

Safe and explainable AI: Uncertainty quantification, chain of thought monitoring
RL, so models can learn from real-world experimentation and unlock superhuman capability
VLM’s inspired from human psychology: system 1/2 thinking, emotionality
Physics and geometry-inspired models, from NeRFs to SE(3) or O(3) equivariance

Applications

Biomedicine: e.g. long-horizon cancer diagnosis from medical imaging
Robotics: e.g. crisis response, closed-loop drug discovery
Fusion: improved control and simulation of fusion devices
Satellite Imagery: e.g. aid human rights investigations by automating large-scale search of objects, using rich temporal, multispectral imagery </aside>

<aside> 🇺🇸

How can the benefits of an automated “AI world” help the working class and poor?
- Funding AI research that is socially beneficial but not immediately or massively profitable, i.e. AI for accelerated science, education, and humanitarian goals
- How do we gracefully deal with jobs lost due to automation, given that net economic output will be great?
Reducing US traffic fatalities, which are roughly equal to the number of gun fatalities (~40K)
- Double the amount of public transit, trains when possible and buses otherwise
- Legal requirements for all new cars to have automated driver assistance (ADAS) or hardware speed limiters, such as Federal Motor Vehicle Safety Standard (FMVSS) No. 127
Ensuring the US military and diplomacy apparatus truly reduces suffering at home and abroad
Government transparency, freedom of the press
More voting </aside>