Kartikeyan Subramanyam

Current Research and Work

      Following a presentation of CRASHCART to ARPA-H Director Dr. Alicia Jackson, the system has received active encouragement toward commercialization from multiple government agencies and healthcare professionals. Entrepreneurial efforts to bring CRASHCART to market are ongoing.
    

Computational and Data Sciences Research Specialist

June 2024 – Present

UCSD Center for Healthcare Cybersecurity

Healthcare Ransomware Resiliency and Response Program (H-R3P) · ARPA-H DIGIHEALS

Lead engineer across two concurrent research programs spanning disaster-recovery infrastructure and national-scale hospital network intelligence, delivering systems across both simulation and live hospital environments.
Presented research outcomes to ARPA-H Director Dr. Alicia Jackson; received active institutional encouragement from multiple government agencies to pursue commercialization.

CRASHCART Principal Network Systems Engineer

Designed and deployed a self-contained "hospital-network-in-a-box" to restore critical clinical functionality during ransomware-induced outages, operating fully independent of compromised hospital infrastructure to eliminate co-infection risk during recovery.
Architected a hybrid network (Starlink + Cellular Backhaul) with segmented L2/L3 design, point-to-point wireless links, and private CBRS-based cellular LAN, providing resilient, rapidly deployable connectivity under degraded conditions.
Reconstructed clinical system dependencies (EHR, lab systems, patient monitoring, radiology/PACS, printing) into a portable architecture; solving hidden middleware and integration constraints required for end-to-end clinical functionality.
Engineered critical workflow integrations: remote printing via IPSec VPN tunneling, lab result routing through LIS/middleware stack into EPIC EHR (resource limited), isolated life-critical patient monitoring networks (inherent by vendor design), PACS/DICOM server configuration, and cellular handoff stabilization via PTP master clock (initially restricted by vendor design).
Led first live field deployment at El Centro Regional Medical Center, CA and subsequent deployment at ValleyWise Health Medical Center, AZ: identifying real-world constraints (thermal degradation at 110°F, PoE power delivery failure, physical access limitations) and adapting system design to ensure reliable operation under field conditions.
Led multi-level research team (undergraduate and graduate students) through concurrent research/engineering sprint cycles: initally scoping workstreams and experiments (under guidance from mentors), eventually managing collaborations, and maintaining delivery quality under high-ambiguity, time-constrained conditions.

Ransomwhere? Data Scientist Platform Engineer

Inherited a partially functional system mid-project with no handover documentation; rebuilt it from the ground up while keeping it operational, with no prior experience in any of the relevant tools.
Built an end-to-end automated data pipeline (Raw Data → InfluxDB → Grafana) ingesting and normalizing high-frequency heterogeneous signals (IP probing results, FHIR endpoint health, and NTP beacon traffic) across distributed hospital systems across the US.
Designed tag-and-field schemas in InfluxDB for three structurally incompatible data streams; synchronized data across multiple purpose-specific servers (scanning, backup, consolidation, database) via automated rsync pipelines on hourly cron cycles.
Designed system behavior to handle missing, delayed, and conflicting signals, ensuring downstream inference remained stable despite the absence of reliable ground truth across any single stream.
Built geospatial dashboards in Grafana with complex query logic and transformations to visualize near real-time infrastructure state and surface outage patterns across hospitals at national scale.

Prior Research & Experience

Research Assistant

June 2022 – February 2024

University of California, San Diego

Aviation Network Security (Aerosec Group)

Reverse-engineered commercial aircraft network architecture — reconstructed full multi-VLAN topology, firewall rules, and segmentation logic from selectively available system logs and disk images using Python and Bash.
Mapped all open ports to running processes; identified structural attack surfaces and segmentation weaknesses in safety-critical aviation network infrastructure.
Work is restricted from public disclosure for reasons of national and industrial security.

FHEmem — In-Memory Accelerator for Fully Homomorphic Encryption (SeeLab)

Configured and extended the Microsoft SEAL library for encrypted CNN workload execution under CKKS and BFV encryption schemes.
Modeled and optimized encrypted data flow within CNN architecture to minimize homomorphic operation count; identified computational bottlenecks per layer.
Produced a 20,000-operation benchmark suite (C#, C++) used as primary validation workload for the research paper.

VMaaS Intern — Vulnerability Management as a Service

July 2022 – December 2022

Securin · Albuquerque, NM

Developed NLP-based vulnerability classification model trained on 210,000+ CVE records from NVD — 92% prediction accuracy enabling automated threat prioritization at scale.
Executed vulnerability assessments across enterprise client environments using Qualys, Nessus, Nexpose, and Ivanti Neurons; conducted penetration testing and reverse engineering.
Executed full vulnerability management lifecycle: identification, evaluation, remediation validation, and executive reporting.

Cybersecurity Intern

July 2021 – September 2021

Spire Solutions · Dubai, UAE

Deployed and calibrated enterprise security platforms (CyberObserver, Elastic SIEM, AppSpider DAST, InsightVM) for endpoint analytics and dynamic application security testing across client infrastructure.
Incorporated and calibrated security solutions to clients' existing IT infrastructure across UAE enterprise environments.

Relevant Coursework & Projects

Computer Security

Buffer overflows, side-channel (memory & time), SQL injection, XSS, CSRF
Network attacks: social engineering, reverse engineering, netcat, nmap, tcpdump
Discovered a previously undiscovered vulnerability in the assignment infrastructure allowing access to other students' submissions

Computer Networks

Router — ARP, TCP, UDP, ICMP, longest prefix match, ARP caching (C)
Sliding Window Protocol — selective retransmission, cumulative ACK, reliable data transmission (C)

Networked Services

SurfStore — Dropbox-like cloud file storage with gRPC (Go)
TritonHTTP — HTTP/1.1 web server with pipelining support (Go)

Operating Systems

Nachos OS — threading, scheduling, and multiprogramming from scratch (Java)

Systems Architecture

Yodie — custom processor for FEC tasks; 9-bit fixed-length ISA (SystemVerilog + Python)

Deep Learning

Taxi Travel Time Prediction — MLP Regressor with feature engineering for rush-hour modeling; outperformed LSTM and LightGBM baselines (Python)

Computer Vision

Photometric stereo, edge & corner detection, bag-of-words image classification
CNN for MNIST — 98% test accuracy; SGD vs. Adam experimentation (Python)

Compiler Construction

'snek' Compiler — Rust compiler generating x86-64 assembly; heap allocation, garbage collection, full expression support

Software Engineering

Birds of a Feather — Android app using Nearby API for class-based friend discovery; Agile/Scrum methodology (Java)

Advanced Data Structures

DNS Blocker — hash table-based in-memory database (C)
Game of Life — fastest implementation in a class of ~250 students (ARM-32)
FP Converter, ETL Program (C, ARM-32)

Digital Systems

6-tap LFSR encrypter, traffic signal controller, alarm clock with day/date tracking (SystemVerilog)

Also Completed

Cryptography
Theory of Computation

Kartikeyan
Subramanyam

Current Research and Work

Publications

Patent

Prior Research & Experience

Technical Skills

Education

Relevant Coursework & Projects

Awards & Competitions

Miscellaneous

Clubs & Organizations

Interests