Projects and case studies

Tokenbooks · 2025 to present · Second engineering hire

Tokenbooks payments and financial data

Second engineering hire at Tokenbooks, a fintech platform for crypto and fiat finance. I work across the stack on payments, wallet and exchange sync, accounting workflows, and reporting.

• ·Built payment request intake, approval routing, counterparty and payment details, and the request-to-payment workflow end to end.
• ·Added crypto-to-fiat payments: requestors invoice in fiat while treasury funds payouts in crypto through one controlled, auditable workflow.
• ·Improved the reliability of multi-wallet, multi-chain ingestion with BullMQ async job orchestration, status tracking, and recovery for failed syncs.
• ·Rebuilt sync, payment, and operational error handling with structured, type-safe, user-facing failures.
• ·Built Coinbase OAuth connect and reconnect flows for secure exchange integrations and reauthorization.
• ·Added XLSX exports for accounting and payment workflows, used by finance teams for close and reporting.
• ·Built AI-agent skills and harnesses for guided workflows, deterministic checks, and safer monorepo navigation.

CloudSEK · 2022 to 2024

CloudSEK Platform (XVigil and SVigil)

XVigil and SVigil are CloudSEK's threat intelligence and attack surface monitoring products.

• ·Built data visualization components, dashboards, and new user-facing pages.
• ·Fixed bugs and worked on performance optimization across both products.
• ·Contributed components to CloudSEK's shared in-house design system.

Other CloudSEK work

• ·Built a React report generator used for 2,000+ reports.
• ·Migrated a NestJS backend service to an Nx monorepo.

CloudSEK · 2022 to 2024

Exposure Search

Exposure Search is a public search product on CloudSEK's threat intelligence platform for exposed data tied to a domain or organization.

• ·Co-built the app, which reached 100,000+ searches and generated 3,000+ leads for CloudSEK.

CloudSEK · 2021 to 2022

BeVigil revamp

BeVigil is CloudSEK's mobile security search engine. I worked across its frontend and backend.

• ·Built reusable React and Next.js components for the frontend revamp.
• ·Cut MongoDB count-query time by 88% and cached computed results in Redis.
• ·Built admin dashboards, search features, and filters with React Admin and Sequelize.
• ·Built and improved Node.js APIs.
• ·Built and optimized Docker images for frontend and backend services.
• ·Wrote the SEO strategy adopted for the product.

CloudSEK · BeVigil

Chunked APK upload system

Large APK uploads were slow and limited by file size, so I replaced the single-shot upload with a chunked pipeline.

• ·Cut large-file upload time by 40 to 50% and removed the file-size limit.

MSc thesis code · UNBC · 2024 to 2026

causal-fl-ids-defense

MSc thesis code for causal structure learning in adversarially robust federated intrusion detection.

• ·Raised PGD robustness from 41% to 89.85% at epsilon 0.20, with under 1% clean-accuracy loss and no adversarial training.
• ·Aggregates a learnable sparse 64x64 causal matrix across clients with FedAvg.
• ·Evaluated FGSM and PGD attacks on a 4-client non-IID federation built with PyTorch and Flower.

Research data engineering · UNBC

um-nids-merge-pipeline

Thesis data pipeline for four network intrusion detection datasets.

• ·Merged 4 NIDS datasets into a 109M-row unified corpus with consistent labels and features.