# Suraj Ranganath

Hello. I like to build and do research. Right now, I am working on building and orchestrating agents, AI security, reinforcement learning, EEG tooling, model evals, world-model memory, and data systems.

I am an M.S. Data Science student at UC San Diego. Previously, I helped build BalkanID Copilot and the Identity Intelligence Graph into production products used by enterprise customers. This summer I am interning at C1 in San Francisco, CA.

I am open to new roles, collaborations, and opportunities. Reach me at suranganath@ucsd.edu.

## Links

- Website: https://surajr.dev/
- GitHub: https://github.com/suraj-ranganath
- Google Scholar: https://scholar.google.com/citations?user=5SOfSs8AAAAJ&hl=en
- LinkedIn: https://www.linkedin.com/in/suraj-ranganath/
- Medium: https://medium.com/@surajranganath
- Resume: https://surajr.dev/Suraj_Ranganath_Resume.pdf

## Selected Work

### BalkanID Copilot

Production AI agent for identity security over enterprise identity graphs. Work included agentic identity-security workflows, graph-shaped enterprise identity data, evaluation, monitoring, permissions, text-to-Cypher, and action boundaries. Public links: https://docs.balkan.id/balkanid-copilot/getting-started-with-the-copilot and https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/20250342365.

### Show, Don't TELL

Explainable AI-generated text detection with evidence spans. Public links: https://arxiv.org/abs/2605.27921, https://ai-tells.tech/, and https://github.com/ACMCMC/rl-detector.

### StealthRL

Fairness-aware reinforcement-learning red-team loop for AI-text detectors. Public links: https://arxiv.org/abs/2602.08934, https://stealthrl.pages.dev/, and https://github.com/suraj-ranganath/StealthRL.

### EEGPrep

Python port of EEGLAB, the popular toolbox widely used by EEG researchers. Public links: https://github.com/sccn/eegprep and https://github.com/sccn/eegprep/pulls?q=is%3Apr+author%3Asuraj-ranganath+is%3Amerged.

### NAMO and NAMO-D in Marin

Implemented two new optimizers for foundation-model training in Marin, mentored by Al Merose. Public link: https://github.com/marin-community/marin/pull/2966.

### KV Cache Quantization for Self-Forcing Autoregressive Video Generation

Thirty-three-method study for self-forcing autoregressive video generation. Public links: https://arxiv.org/abs/2603.27469 and https://github.com/suraj-ranganath/kv-quant-longhorizon/.

### PIPE-RDF

LLM-assisted pipeline for enterprise RDF benchmark generation. Public links: https://arxiv.org/abs/2602.18497 and https://github.com/suraj-ranganath/PIPE-RDF.

### TrailKarma

Offline-first trail intelligence app with BLE relay and on-device ML. Public links: https://devpost.com/software/trailkarma and https://github.com/ACMCMC/trailkarma.

### Multimodal Learning Analytics Dataset

Paper with EEG, eye tracking, GSR, video, and spatial reasoning data. Public link: https://www.nature.com/articles/s41597-025-04865-5.

### VaultLens

Telegram personal knowledge-base agent for memory, deadlines, digests, and calendar-aware context. Public link: https://github.com/suraj-ranganath/VaultLens.

### Where Bits Matter in World Model Planning

Mixed-bit world-model planning study showing where low precision breaks. Public links: https://arxiv.org/abs/2602.11882 and https://github.com/suraj-ranganath/DINO-MBQuant.

## Timeline

- Jun 2026 - now: C1, incoming software engineering intern, AI-native identity security.
- Jun 2026 - now: Swartz Center for Computational Neuroscience, UC San Diego, graduate software engineer with Dr. Arnaud Delorme.
- Jan 2026 - Apr 2026: Halicioglu Data Science Institute, UC San Diego, graduate software engineer with Dr. Victor Minces.
- Jan 2026 - Feb 2026: Qualcomm Institute, UC San Diego, AI research program mentor.
- Sep 2025 - Jan 2027: UC San Diego, M.S. Data Science, GPA 4.0/4.0.
- May 2023 - Jun 2025: BalkanID, machine learning intern to machine learning engineer.
- Jan 2023 - Jun 2023: IIT Bombay EdTech Lab, research intern.
- Jun 2022 - Sep 2022: Strive Math, YC S21, entrepreneur in residence intern.
- Jul 2022 - Aug 2022: MySivi, YC W22, machine learning intern.
- Aug 2020 - May 2024: Vellore Institute of Technology, B.Tech. Computer Science with specialization in Bioinformatics.

## Patents

- Knowledge Graph-Enhanced AI Copilot Platform for Intelligent Identity Security Governance and Lifecycle Management.
- Recommendation and Remediation of Role-Based Access Control Postures for Identities.
- Multi-Agent Identity Governance and Administration System.
- Method for Provisioning and De-Provisioning Just-in-Time, Purpose-Based Access for Identities within Applications.

## Research

- Show, Don't TELL: Explainable AI-Generated Text Detection.
- KV Cache Quantization for Self-Forcing Video Generation: A 33-Method Empirical Study.
- PIPE-RDF: An LLM-Assisted Pipeline for Enterprise RDF Benchmarking.
- Where Bits Matter in World Model Planning: A Paired Mixed-Bit Study for Efficient Spatial Reasoning.
- StealthRL: Reinforcement Learning Paraphrase Attacks for Multi-Detector Evasion of AI-Text Detectors.
- A Multimodal Dataset of South Asian Post-Graduate Students Working on Mental Rotation Tasks.