Using LLMs to accelerate cybersecurity processes - BC-1004

Project type: Research
Desired discipline(s): Computer science, Mathematical Sciences
Company: Anonymous
Project Length: 4 to 6 months
Preferred start date: As soon as possible.
Language requirement: English
Location(s): BC, Canada
No. of positions: 6 - 12
Desired education level: CollegeUndergraduate/BachelorMaster'sPhDPostdoctoral fellowRecent graduate
Open to applicants registered at an institution outside of Canada: No

About the company: 

Plurilock is a cybersecurity company focused on delivering critical services and defense-oriented solutions to government and enterprise clients. The company operates at the intersection of cybersecurity, national defense, and critical infrastructure, providing services such as security assessments, zero trust implementation, and data protection.

Describe the project.: 

We are exploring collaborations with researchers, postdocs, faculty, and research labs on applied cybersecurity problems at the intersection of AI, autonomous systems, and mission-critical infrastructure. We’re particularly interested in work that combines novel research with real-world validation opportunities.

1) Cyber Assurance for Autonomous Systems

How do we ensure autonomous platforms (UAS, subsea, space) remain secure in contested environments?

Areas of interest:

Formal verification and runtime assurance for autonomous decision-making
Adversarial robustness of onboard AI/ML systems (e.g., sensor spoofing, model manipulation)
Secure architectures for embedded, resource-constrained, and intermittently connected systems
Hardware/firmware trust (secure boot, attestation, anti-tamper)
Red-teaming and validation of cyber-physical systems

2) Autonomous Cyber Resilience (Enterprise & Critical Infrastructure)

How can cyber defense systems dynamically adapt and respond at machine speed?

Areas of interest:

Autonomous “cyber maneuver” (dynamic reconfiguration of systems, identities, and networks)
AI-driven defense orchestration under uncertainty
Self-healing systems (closed-loop detect → respond → validate)
Game-theoretic/adversarial modeling of attacker-defender dynamics
Integration of exposure management, control validation, and SOC operations

3) Vulnerability Research for Mission Systems (Firmware & Embedded)

How do we uncover and mitigate vulnerabilities in opaque, mission-critical technology stacks?

Areas of interest:

Automated firmware analysis and vulnerability discovery
Binary analysis, fuzzing, and symbolic execution for embedded systems
Hardware/software interaction vulnerabilities (side channels, fault injection)
Supply chain integrity and tamper detection
Scalable approaches across heterogeneous device ecosystems
Collaboration Model

We are interested in postdocs and academic labs looking to work on high-impact, dual-use problems with pathways to deployment.

We can support:

Access to real-world systems and problem sets
Alignment with Canadian and allied research funding opportunities
Opportunities for publication + applied validation

4) OSINT, Intelligence Support & Digital Investigations (LLM-Enabled)
How can we leverage LLMs and advanced AI to transform open-source intelligence (OSINT), investigative workflows, and intelligence analysis at scale?
Areas of interest:
• LLM-assisted OSINT collection, triage, and entity resolution across fragmented data sources
• Multi-modal analysis (text, image, video, geospatial) for investigative workflows
• Automated link analysis, knowledge graph construction, and narrative generation
• Detection of deception, synthetic media, and coordinated influence operations
• Human-AI collaboration models for analysts (trust, explainability, and workflow integration)
• Privacy-preserving and legally compliant approaches to large-scale data collection and analysis
Why this is interesting academically:
• Bridges NLP, information retrieval, knowledge representation, and security studies
• Rapidly evolving space with clear gaps in evaluation, robustness, and trustworthiness
• High relevance to national security, law enforcement, and enterprise risk
• Opportunities for both methodological contributions and real-world datasets/use cases

Required expertise/skills: 

LLMs, AI