Security Engineer · AI Researcher

Dinesh
Merges
Security & AI.

Cybersecurity engineer and AI researcher focused on machine learning, deep learning, and securing the internet space for the people — building systems that are both intelligent and trustworthy.

AI Security Deepfake Detection Digital Forensics Automotive CyberSec
Dinesh

Security meets
machine learning.

I'm a cybersecurity researcher with a deep interest in how AI systems can both create and defend against new threat vectors. My work sits at the intersection of computer vision, adversarial ML, and real-world security challenges.

I'm drawn to problems where the stakes are high: synthetic identity fraud, deepfake-enabled disinformation, and the arms race between generative models and detection systems.

🔬

Research

Adversarial ML, deepfake detection & synthetic media forensics

🧠

ML Engineering

PyTorch, transfer learning, GradCAM interpretability, W&B

🛡️

Security

Identity fraud defense, threat modeling, AI robustness

🚀

Deployment

Streamlit, Hugging Face Hub, cloud-native ML pipelines

Projects

GitHub →
🗺️
Geospatial · Research Live Demo

GeoInsight — Research Visualization Platform

Developed in collaboration with Queen's University Belfast, this platform analyzed the impact of pollution-awareness videos shown to children by combining geospatial mapping with data visualization.

  • Built the full web interface using HTML, CSS, and Google Maps API
  • Automated conversion of location data into latitude/longitude coordinates
  • Implemented real-time map updates using a secured Google API key
  • Generated analytical graphs to visualize behavioral impact metrics
  • Integrated data workflows using Google Colab for processing and analysis

Led all technical development, delivering a live, interactive system that mapped video deployment locations and visually represented research findings through dynamic charts and geospatial overlays.

HTML/CSS Google Maps API Google Colab Data Visualization Geospatial
View Live →
⚙️
Security Research

More Coming Soon

Additional projects in adversarial ML and AI security tooling currently in progress. Check back or follow on GitHub.

In Progress
Technical Stack

Skills & Tools

Tech Stack
  • Go
  • Python
  • C / C++
  • HTML / CSS / JavaScript
  • Bash / Shell Scripting
Cybersecurity
  • Penetration Testing
  • CAN Bus & ECU Security
  • Threat Modelling
  • Cryptography
  • Network Security
  • Deepfake Detection
  • Digital Forensics
  • Fault Injection Testing
Dev & Research Tools
  • GitHub
  • Google Colab
  • Jira
  • Linux
  • Wireshark
  • VS Code
  • Google Maps API
  • Streamlit
AI Tools & Platforms
  • Claude (Anthropic)
  • ChatGPT (OpenAI)
  • GitHub Copilot
  • Gemini (Google)
  • Hugging Face
  • Weights & Biases
  • Midjourney
  • Stable Diffusion
Background

Experience &
Education

2025 – Present

MSc Applied Cybersecurity

Queen's University Belfast · Belfast, UK

Modules: Network Security  ·  Pentesting & Ethical Hacking  ·  Cryptography  ·  Cyber in AI  ·  Software Assurance

May 2022 – Sept 2025

Project Engineer

Wipro · Bengaluru, India
  • Performed integration and CAN network security testing for Mercedes‑Benz Truck Instrument Clusters, ensuring secure ECU communication and compliance with automotive cybersecurity standards.
  • Analyzed CAN bus traffic, detected anomalies, and validated message integrity and ECU authentication mechanisms.
  • Conducted stress, high-load, and fault-injection tests to assess system resilience against abnormal or malicious inputs.
  • Collaborated with global suppliers to review security requirements and strengthen cross-module communication safeguards.
  • Automated secure ECU flashing workflows, improving efficiency and reducing human error in software updates.

Let's connect.

Always happy to talk AI security, research, or anything at the intersection of machine learning and adversarial systems. Reach out through any of the links below.

If you have any cool and crazy ideas to work on, just ping and let's get started.

"Sounding the alarm before the breach — that's the only defense that matters."

— Dinesh Sangwan · Applied Cybersecurity Researcher