AI observability and shadow AI discovery for dynamic governance - ON-1256

Project type: Research
Desired discipline(s): Engineering - computer / electrical, Engineering, Computer science, Mathematical Sciences, Statistics / Actuarial sciences
Company: FairFuture AI Inc.
Project Length: 6 months to 1 year
Preferred start date: 01/04/2027
Language requirement: English
Location(s): Toronto, ON, Canada
No. of positions: 1
Desired education level: Master'sPhDPostdoctoral fellowRecent graduate
Open to applicants registered at an institution outside of Canada: No

About the company: 

FairFuture AI is a Canadian company focused on Responsible AI research and development and building an enterprise AI governance platform. The platform is designed to help organizations maintain an AI inventory, assess risk, manage controls and evidence, and adapt oversight as models, data, policies and AI agent behaviour change. Fiara is the intelligence layer intended to simplify this governance experience.

Our ambition is to develop a global Responsible AI platform and a Canadian research hub through sustained academic and industry collaboration. These three projects address immediate commercialization and technical research priorities within that longer-term program. We aim to create commercially useful intellectual property in Canada, develop skilled Canadian talent, support future employment and build export opportunities. Research outputs are intended for incorporation into FairFuture's platform, subject to agreed academic IP, licensing and publication arrangements.

Describe the project.: 

Dynamic AI governance requires knowing which AI systems and agents are operating, how they relate to registered use cases, and when governance evidence becomes incomplete or outdated. This Accelerate project will investigate how heterogeneous, partial observations can reconstruct an initial governance state and continuously identify changes, including unregistered AI activity. The research question is whether confidence-aware evidence fusion improves discovery and state reconstruction over inventory matching and isolated telemetry rules.

The intern will study entity resolution and provenance across three bounded sources: instrumented agent traces, model/API gateway records and cloud/software inventory metadata. An authorized test environment will contain known registered and deliberately unregistered AI assets. Activity outside monitored sources will be treated as a coverage limitation rather than claimed to be discoverable.

Methods will include a literature review, temporal evidence graphs, probabilistic matching and anomaly/change detection. Experiments will compare exact identifiers, rule-based correlation and confidence-aware fusion using ground-truth datasets with missing, delayed, conflicting and duplicated events. Evaluation will measure discovery precision/recall, identity-linking accuracy, change-detection delay, confidence calibration, provenance completeness and processing overhead. Evidence records will distinguish observed, inferred, stale and unknown states; minimization and redaction will reduce sensitive content collection.

Months 1–2: evidence schema, research hypotheses, coverage assumptions and test dataset. Months 3–4: discovery and reconciliation prototypes. Months 5–6: controlled evaluations and robustness studies. Months 7–8: integration and reproducible validation.

Outputs will include reusable discovery/correlation components, an evidence model, benchmark datasets, change events and APIs connecting to FairFuture's AI Registry and governance engine. These outputs will inform supervisory agents while remaining independently testable. Weekly academic and company supervision will support research and integration. The project targets uncertainty in governance evidence rather than routine dashboard construction and aims to create Canadian IP and skilled research capacity, with platform-use rights agreed before work starts.

Required expertise/skills: 

Required: Python, SQL and data pipelines; telemetry and distributed tracing; entity resolution or data integration; probabilistic modelling, anomaly detection or temporal graphs; experimental design and statistical evaluation. Ability to create ground-truth datasets, quantify uncertainty and document observability coverage is essential.

Preferred: a research master's or PhD student or recent graduate in computer science, statistics or computer engineering. Assets include OpenTelemetry, knowledge graphs, event-driven systems, AI/LLM observability, cloud inventories, security analytics and privacy-preserving telemetry. Containerized implementation, PostgreSQL and provider-independent APIs will support platform transfer. Candidates should distinguish evidence of AI activity from assumptions about ownership, authorization and risk. A suitable academic supervisor would bring expertise in data integration, observability, probabilistic systems or security; FairFuture's technical lead will support engineering integration. Dashboard construction or connector implementation alone is insufficient for this research scope.