Engineering student at UC Berkeley.
I study EECS and Industrial Engineering & Operations Research. I’m exploring AI safety evaluations, token-efficient AI systems, and prediction markets.
What I’m working on ↗UC Berkeley Haas School of Business · Undergraduate researcher
I develop Python and PyTorch pipelines for multimodal behavioral analysis, extracting facial, vocal, and paralinguistic features from video and audio. These measures support the lab’s research on political ideology, race, and socioeconomic status.
UC Berkeley AI Research · Student researcher
I built a 1,020-prompt benchmark to study bias in model decisions across healthcare, criminal justice, and business scenarios. I collected model choices and logits from FLAN, Vicuna, LLaMA, and LLaMA 2.
Read the resulting paper ↗A benchmark examining model decisions across race, gender, age, socioeconomic status, and political bias.
Read on IEEE Xplore ↗Exploring how model routing, caching, and context management affect cost and answer quality. Includes a literature review and two local prototypes; live-model evaluation is next.
View project on GitHub ↗I built a system that turns legal agreements into executable state machines for validation and scenario testing. Multiple agents share context and use validation feedback to refine the output.
View Babel on GitHub ↗An inbox agent router that starts with small local models and escalates when needed. I benchmarked latency and token usage across Ollama, OpenAI, and Claude.
View SLM Cascade on GitHub ↗UC Berkeley’s student-run rocketry team. Students design, build, and test rockets, including liquid-propellant vehicles.
I built C++ support for an environmental sensor board, parsing UDP packets for temperature, humidity, and pressure telemetry. I integrated the board into STAR’s configuration, data streams, and monitoring interface.
B.S. in Electrical Engineering & Computer Science and Industrial Engineering & Operations Research
Built C++ support for environmental sensor telemetry, from UDP packet parsing to the data acquisition interface.
Built assistive AI for facial emotion recognition. Interviewed 50+ clinicians and users and secured partnerships with three autism clinics.
Created a balanced dataset of 40,000+ videos and images to reduce bias in an emotion-recognition model. Integrated the model into robotics applications.
Python · C++ · C · Java · PyTorch · Git · Ollama · Kubernetes
Research roles are listed in Research. The PDF includes the full details.
Updated September 28, 2026
This portfolio’s code does not include advertising, analytics, tracking cookies, or a contact form. It saves your light or dark theme choice in your browser’s local storage. You can remove that preference by clearing this site’s browser data.
The public website is hosted on GitHub Pages. GitHub logs visitor IP addresses for security purposes. See GitHub’s privacy statement.
A separate preview is hosted through OpenAI Sites on Cloudflare. Its hosting and sign-in services may process technical information, such as IP addresses, browser details, and access logs.
Spotify and other external links open separate websites with their own privacy policies. No Spotify player is embedded here. If you email me, I receive your address and the information you choose to send through my email provider.
For privacy questions, email pranay.dogra@berkeley.edu.
The site supports keyboard navigation, light and dark themes, and your device’s reduced-motion setting. ASCII animations play briefly, then stop. The resume summary is available as text alongside the PDF.
If something is difficult to use, email pranay.dogra@berkeley.edu.