Public sector advisory

Pro bono consulting work: non-profits and public service

Munk School International Institute on Social Policy

Information Management Branch• Assessed cybersecurity risks of the emergent technologies, suggesting MLOps solutions (edge AI, RAG models, adversarial machine learning, quantum-as-a-service). • Aggregated cost-effective practices in digital transformation for at-risk environments (innovation sandboxes, rotational quality assurance, business value segmentation, matrix organization).

Munk School of Global Affairs & Public Policy• Preparing Mitacs-funded policy brief on Ukraine’s gun policy, regulation incidence, and risk in this area for a case competition; to be pitched to the Ukrainian/Canadian stakeholders.• Proposed supply-incidence mechanisms for firearm black-market seizure and regulatory compliance—declarations of conformity, intermediary agency, and buyback amnesty program.• Performed jurisdictional scans of gun control policies of Estonia, Georgia, Northern Ireland (UK), and New Zealand.

Centre for Criminology and Sociolegal Studies (CrimSL)• Collected data on stakeholders via respective bills, collective addresses, and governmental decrees and scrubbed it in R by thresholding.• Conducted exponential random graph model (ERGM) analysis with three significant factors predicting coalition-building between select MPs.• Led a team to perform a SWOT analysis of current legislative proposals and create an innovation index for select MPs (comprised of voting deviation, sessional attendance, and percentage of pledges fulfilled).

• Parsed, cleansed, and tokenized over 120,000 messages on Telegram in Python. • Calculated cosine similarity scores between 12 select Russian-speaking war correspondents and assessed the significance of their differences in a 95% confidence interval using the two-sigma rule, Barlett’s sphericity test, and network analysis tools.• Trained a Latent Dirichlet allocation (LDA) model that extracted keywords from 120,000 messages and constructed the top 5 most occurring topics, observing risks in what field the industry is most likely to react to.• Executed a variance change-point analysis of time series data for the number of subscribers on 12 Telegram channels, finding major nine 'waves' of change points over 23 months.

• Using the process-tracing case study method and analyzing over 70 primary sources, outlined a stakeholder map of 16 state and non-state actors advocating for gun control liberalization.• Found that the Ukrainian hybrid regime is moderately conducive to the aspirations for private gun ownership, with social cohesion disruption and administrative capacity ambiguity as potential drivers.

• Helped extract app. 2,800 paragraphs in Python to train a natural language processing model snowballing key terms and generating the most popular contexts of the scraped Russian-language articles about security and foreign policy. • Assisted in automating large-number searches and summarizing the meanings of select terms in Russian strategic decision-making (>1,000 scholarly articles and commentaries).• Engaged in teamwork-based documentation of protocols, findings, and limitations of the code and procedures enacted.

• Processed and merged seven waves of Ukrainian surveys and European Social Survey (ESS) datasets using R to generate dummy variables for belief network analysis.• Coded historical data spanning over a century to track state capacity and economic welfare dynamics in Ukraine's local governance units.

Wrote news digests and explainers; investigated popular fakes and narratives of disinformation; conducted fact-checking of Ukrainian politicians.

Made transcriptions of focus groups and interview recordings; coded TV shows and Twitter and Facebook pages of “Donetsk People’s Republic” and “Luhansk People’s Republic” for content analysis via NVivo 10.