Trinh Van Thuong (Jasmine)
System Architect & Business AnalystSystem Architect with experience across fintech, payments, consumer finance, IoT operations, and enterprise AI platforms. I turn product and operational needs into clear specifications — covering requirement discovery, process mapping, solution design, API and integration flows, data rules, access control, audit trails, reconciliation, and delivery support.
Working with business users, compliance teams, vendors, and engineers to translate operational and product needs into systems that hold up in production.
Selected platforms across regulated finance, payments, security operations, and enterprise AI — from target architecture and requirement scope through delivery support.
Knowledge retrieval, workflow automation, and operational analytics. Built RAG pipelines for document ingestion, chunking, hybrid retrieval, reranking, multilingual translation, and citation-backed responses — with multi-tenant edition isolation, role/tier access control, OAuth connectors, semantic caching, and full observability.
Defined target architecture and requirement scope for a regulated remittance platform: customer onboarding, transaction processing, compliance checks, and reconciliation. Mapped transaction states, settlement flow, exceptions, and disputes; designed modular eKYC and risk-check components so policy rules could change without touching core processing.
A shared platform for physical security, IoT access control, device events, monitoring, and operational workflows. Clarified user roles, access rules, approval flows, and device lifecycle behavior; used event-based processing for high-frequency access and monitoring events while keeping device logic separate from core services.
Designed the integration layer between core financial systems and a customer engagement platform. Worked through data-exposure rules, customer event definitions, consent boundaries, and access limits — enabling controlled event propagation so engagement features could use behavior data without exposing sensitive core financial data.
A modular financing platform for loan products, repayment flows, customer status, fees, approval steps, and operational controls. Mapped financial state changes so loan events, account updates, repayments, adjustments, and exceptions could be traced and audited, with access-control rules separating business, operations, and system responsibilities.
Defined payment, subscription billing, validation, transaction status, and operational support flows. Designed event-based reconciliation between payment status and physical access control so service access updated on payment results, and prepared API/integration requirements covering failed payments, delayed confirmation, and reconciliation edge cases.
A multi-brand dealer group moving from fragmented per-dealership sites to one omnichannel platform: browse group-wide stock, reserve with a refundable deposit, pre-qualify for finance, value a trade-in, and book a test-drive or service — online, 24/7. End-to-end BA/SA artifacts plus interactive wireframes.
View case studyThis case study is under active development — content, artifacts, and wireframes are still being refined.
A worked BA/SA case study for a multi-brand dealer group moving from fragmented per-dealership sites to a single omnichannel platform: browse group-wide stock → reserve with a refundable deposit → pre-qualify for finance → value a trade-in → book a test-drive or service — online, 24/7.
7 dealerships across 4 OEM brands plus a used-car division, each on a separate ageing site. Customers can't transact online; enquiries land in shared inboxes with a ~9h median first response and ~30% never logged in the CRM; group stock isn't visible across locations; service bays are phone-only; and finance-advertising disclosures are inconsistent.
In: group inventory catalogue & search, vehicle detail with finance calculator, reserve-with-deposit, finance pre-qualification (soft-pull), trade-in valuation, test-drive & service booking, lead routing, dealer admin, consent & disclosure.
Out (this release): full online purchase / e-signature, home delivery, parts e-commerce, OEM build-and-price configurator, loyalty programme.
Browse the full documentation set — 16 artifacts across five delivery phases — from the menu on the right.
From requirement to running system — I translate business intent into specifications engineers can ship, designing flows, data, and integrations that hold up under real load, where strategy sets the direction and engineering proves it in production.
Analysis and system-design practice grounded in concrete tooling across data, backend, cloud, and AI platforms.
Verified specializations behind the work — kept current to sharpen the engineering and architecture fundamentals. Select a certificate to view it full size.
Occasional posts from my LinkedIn — on career, systems thinking, and the craft of building software that lasts.
Speed should never replace judgment — why copying AI output without review is a habit to break, and how asking “Would I put my name on this?” keeps human accountability central.
Read on LinkedIn
Great solutions begin with understanding the real goal — why even the best system fails without a meaningful business problem, and how outcome-first questions align the tech.
Read on LinkedIn
Consistent communication beats constant meetings — why the best teams aren't the ones that meet most, but the ones with a predictable rhythm that turns communication into a system.
Read on LinkedIn
The most powerful tools amplify human potential — why frontier AI isn't replacing people but acting as a flywheel, with human judgment still the differentiator.
Read on LinkedIn
Sometimes a walk in the park is worth more than another hour online — why health, energy, and time to step away from the screen are a kind of wealth too.
Read on LinkedIn
Lessons from AI 2041 by Kai-Fu Lee — why the future of AI is as much a human problem (privacy, ethics, jobs) as a technology one, and why preparing for it starts today.
Read on LinkedIn
Building a sustainable career without sacrificing what matters most — why balance isn't splitting time equally, but having enough energy to show up fully for both.
Read on LinkedInThe things hiring managers, product owners and clients usually want to know before a first call — answered straight from the work on my CV.
I sit between the business and the build. The analysis half is requirement discovery and clarification, process mapping and solution design — then writing the functional specifications, user flows, process diagrams, data rules, API behavior, integration flows, access rules and acceptance criteria the team builds against.
The architecture half is designing backend modules, partner integrations, event flows, transaction states, reconciliation logic and the audit trails and reporting data operations and compliance teams need — plus the calls on service boundaries, database usage, security permissions, deployment constraints and non-functional requirements.
I stay with it through delivery: coordinating product owners, business teams, developers, QA, infrastructure and external vendors from requirement review through testing and release, then reviewing implementation gaps and adjusting the design when business rules, regulations or operational needs change.
Fintech, payments, remittance, consumer finance, subscription billing, regulated systems, IoT access control and enterprise AI workspaces. I have been with Fintech Innovation Solution Joint Stock Company since 2022, as Business Analyst and System Architect.
The selected platforms are the Enterprise AI Platform (2026), Cross-Border Remittance Platform (2025), Facility Management & Security Platform (2025), Customer Engagement & Analytics Integration (2024), Consumer Financing Management System (2023) and the Digital Payment & Subscription Management Platform (2022).
Traceability is a design input, not paperwork bolted on at the end. On the cross-border remittance platform I defined the audit trail, traceability and reconciliation requirements that let compliance teams review and investigate individual transactions, and mapped transaction states, settlement flow, exception handling and dispute handling for both business and operations teams.
I also keep policy separate from core processing — eKYC and risk-check components were designed as modules so policy rules could change without touching core transaction processing. The same principle applied in consumer financing: financial state changes were modelled so loan events, account updates, repayments, adjustments and exceptions could all be traced and audited, with access-control rules separating business, operations and system responsibilities.
The Enterprise AI Platform — knowledge retrieval, workflow automation and operational analytics. That included RAG pipelines covering document ingestion, chunking, hybrid retrieval, reranking, multilingual translation and citation-backed responses, plus Gmail-to-task ingestion, meeting transcription and summarization, dashboard analytics, file generation and SSE chat streaming.
Architecturally: multi-tenant edition isolation, role and tier access control, OAuth connectors, semantic caching, and monitoring through Prometheus, Grafana and Langfuse. The AI tooling is Claude, Ollama, LiteLLM and Qdrant.
Data and backend: PostgreSQL, SQL Server, SQLite, MongoDB, Redis, Qdrant, Python, FastAPI, Node.js, TypeScript, Next.js and Docker. Cloud and delivery: AWS, GCP, hybrid cloud and on-premise systems, OAuth connectors, SSE, Prometheus, Grafana and Langfuse.
I work in Agile, Scrum or Waterfall depending on the client, and run delivery through ClickUp, Trello, Jira and Notion — task tracking, sprint planning, backlog management and release coordination, with UAT support and stakeholder coordination on the analysis side.
Background: Bachelor of Software Engineering, Ton Duc Thang University (2018–2022). I work in Vietnamese (native) and English (professional working proficiency).