RaynaVB (Rayna Budigelli) · GitHub
Skip to content
View RaynaVB's full-sized avatar

Highlights

  • Pro

Block or report RaynaVB

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
RaynaVB/README.md

Rayna Budigelli

AI Engineering (BSE) + Wharton @ Penn · Building Varma LLC · Shipped Plateify to the App Store

I build data products end to end: model, backend, interface, ship. Most of my work sits where machine learning meets a real decision someone has to make.

Currently recruiting for Summer 2027 SWE / ML / Data internships.


What I'm working on now/Future Projects

  • Plateify: maintaining the live iOS app; Android port in progress
  • Allergen menu scanner: AI that flags likely-safe dishes from restaurant menus; exploring a path to restaurant-verified data
  • Coursework: statistical computing in R (STAT 4050), engineering statistics (ESE 4020), dynamical systems (ESE 2100)
  • Reading: human–AI decision-making and causal inference, ahead of a Spring 2027 research assistantship

Projects

Plateify: iOS · Live on the App Store

Connects what you eat to your mood and energy. No calorie counting.

Photo of a meal → Google Vision extracts ingredients → a pattern engine correlates foods against the user's own check-ins. The interesting part is the signal problem: the engine upweights infrequently-eaten foods that track with how someone feels and downweights foods that show up on good and bad days alike, then lets users run structured elimination tests to confirm what it surfaced.

Swift · Firebase · Google Vision AI · pattern detection

AgentDeck: Desktop LLM agent with approval gates

An Electron app hosting a Claude agent with four local tools: read file, write file, run shell, fetch URL. Built around a security boundary — the renderer has no filesystem or shell access, the main process holds the API key, and every destructive operation goes through an explicit user approval gate.

Electron · JavaScript · Anthropic SDK · IPC · tool use

Email engagement analysis: Built In (internship)

Analyzed user email engagement in Snowflake to isolate the drivers behind an engagement drop, and turned the result into a recommendation the team could act on.

SQL · Snowflake · analytics


Stack

Languages: Python · SQL · Swift · JavaScript · R · Java

Data: Snowflake · pandas · statistical computing in R

Build:Firebase · Electron · Google Vision AI · Git · Claude Code

Interests: human–AI decision-making · causal inference · experimentation


Experience

Founder & Engineer Varma LLC: solo-building and shipping consumer apps
Teaching Assistant CIS 1210, Data Structures & Algorithms @ Penn
Data/Analytics Intern Built In

Reach me

LinkedIn Email Website

Open to: Summer 2027 internships (SWE, ML, data) · research collaborations in human–AI decision-making


Penn '29, BSE in Artificial Intelligence + Wharton (uncoordinated dual degree)

Popular repositories Loading

  1. agentdeck agentdeck Public

    A minimal desktop agent harness. Chat with Claude, let it call local tools, approve every side-effectful action.

    JavaScript

  2. RaynaVB RaynaVB Public