劉岦崱 · Nils — GenAI Solution Architect
Bringing AI Agent Platforms
to Financial Institutions
That Actually Work.
Making AI Agents genuinely usable inside regulated financial environments — compliant, auditable, and production-ready.
I'm Nils (劉岦崱) — the person who builds AI Agent platforms that actually work inside financial institutions. Currently leading an overseas AI team at a major Taiwanese financial holding, building AI products from 0 to 1 with a cross-functional team. I built EnterpriseGPT (the bank's first production GenAI platform), AI Coach (96% satisfaction across 100+ financial advisors), and the Overseas AI Agent Platform (3-layer Agent architecture, deployed in Vietnam). I hold 6 utility patents, all in LLM application system architecture. My focus is production, not demo: regulated environments, compliance constraints, and real users in real financial institutions.
Impact in Numbers
Not a feature list. Real outcomes shipped in production.
Global Trajectory
NTU — B.Sc. Chemistry
Graduated from National Taiwan University, building a rigorous scientific foundation.
TU Chemnitz — M.Sc. Chemistry
Pursued a Master's degree in Germany, gaining international academic and cross-cultural experience.
Groningen Univ. Visiting Researcher / Fraunhofer Intern
Conducted advanced materials and nano-synthesis research at a Nobel Laureate lab (Holland) and Fraunhofer (Germany).
Avancis GmbH — Int'l Project Specialist / PM (Germany)
Managed tech-transfer projects and pilot-line implementations, coordinating cross-border teams for automation acceptance.
DBS Bank — MA → DevOps Sr. Associate
Joined via Management Associate program. Led Python automation and ML enablement for 200+ staff.
Synpulse — Banking Tech Consultant
Consulted at a Swiss financial advisory firm; built the GenAI risk analytics module 'Narrator'.
Leading Financial Holding — Overseas AI Team Lead
Leading overseas AI team building from 0→1. Architect of EnterpriseGPT (the bank's first production GenAI), AI Coach (96% satisfaction), and the Overseas AI Agent Platform (Vietnam). 6 patents granted.
Featured Prototypes
See how I build production-grade AI Agents — from concept to working demo.
AI KYC Agent — Autonomous Browser Automation
An AI Agent that controls the browser like a human to perform KYC verification — fully autonomous, fully audited.
Autonomous Agent
AI controls the browser like a human operator — zero manual intervention required.
Full Audit Trail
Every action is recorded with screenshots and structured audit logs for compliance.
Auto-Generated Report
A structured KYC report is automatically produced upon task completion.
Compliance-Ready
Built for regulated environments — governance and observability from day one.
News Sentiment Agent Team — Multi-Agent Research
An AI Agent Team that breaks language and time barriers — users discuss news with AI, which automatically collects, summarizes, cites sources, and exports or emails reports.
Multi-Agent Collaboration
Agent team members work together and can be infinitely expanded for new capabilities.
Cross-Language Research
Break language and timezone barriers — discover and analyze global news in any language.
Interactive Q&A
Chat with AI to drill down into stories, uncover related news, and ask follow-up questions.
Auto Summary & Export
AI collects info, summarizes with sources attached, and exports or emails the final report.
Insurance Life — Into the Mist (Lead Dev: Yen-Min Chiu)
A life simulation system that visualizes risk management and financial planning, mapping life's unexpected events to actual insurance claim logic.
Risk Visualization
Pre-warns of life events (like accidents or illness) so players can plan their safety nets in advance.
Actuarial Mapping
Dynamically maps unexpected events to real-world healthcare and insurance payout results.
Dynamic Age Engine
Accurately simulates changing income, family responsibilities, and risk probabilities over time.
Interactive Financial Education
Breaks down jargon barriers to provide an interactive experience with real-world claim logic.
AI Interview Assistant — Live Google Meet Participant (Lead Developer: Sunny Sun)
An AI Agent that joins Google Meet as a named participant, providing real-time interview support and automated meeting documentation.
Live Meeting Participation
AI joins Google Meet directly as a named participant — no plugins or screen capture needed.
Interview Workflow Support
Assists interviewers in real-time with structured scoring and summary feedback on candidates.
Auto Meeting Minutes
Automatically records the full conversation and generates exportable structured meeting summaries.
Multi-Interviewer Integration
Supports multiple interviewers simultaneously, synthesizing different perspectives and evaluations.
Services
Focused engagements designed to move fast and deliver measurable outcomes.
Consulting
I help leadership teams cut through AI hype and make confident, informed decisions. Expect a clear roadmap, not a slide deck.
Best for: strategy clarity, architecture review, build-or-buy decisions
AI strategy assessment, architecture blueprint, vendor evaluation, risk & governance framework, executive readout.
Rapid Prototyping
I build production-grade prototypes that prove what's possible — and what isn't. Your team inherits clean code and clear documentation.
Best for: proving value fast, de-risking before committing headcount
Working prototype, evaluation suite, cost model, handoff documentation, demo recording.
Enablement & Training
I run intensive, hands-on workshops tailored to your team's stack and skill level. No generic content — everything is built around your real problems.
Best for: upskilling engineering teams, internal AI champions
Custom workshop, hands-on labs, reference materials, follow-up Q&A session.
6 AI Patents — The Stories Behind Them
Sole inventor on all six. Each patent originated from a real enterprise problem, not an academic exercise.
Financial Investment Advice Generation System
LLMs produced unstructured, unauditable investment recommendations. This patent enforces structured output that meets financial compliance requirements.
Intelligent Optimization System
Enterprise prompts were managed ad-hoc with no versioning. This patent establishes a version-controlled prompt optimization mechanism.
Information Query System
Traditional keyword search fails on unstructured enterprise documents. This patent introduces a semantic retrieval architecture for accurate information access.
Customizable Generative AI System
Different bank departments needed custom AI behaviors on the same underlying model. This patent solves the multi-scenario deployment scaling problem.
Data Query System
Cross-system data queries were fragmented and costly to integrate. This patent provides a unified query interface to reduce engineering overhead.
Knowledge Graph Construction System
Unstructured enterprise knowledge was inaccessible to AI systems. This patent builds a reasoning-capable knowledge structure so AI genuinely understands business context.
About
I'm Nils (劉岦崱) — the person who builds AI Agent platforms that actually work inside financial institutions. Currently leading an overseas AI team at a major Taiwanese financial holding, building AI products from 0 to 1 with a cross-functional team. I built EnterpriseGPT (the bank's first production GenAI platform), AI Coach (96% satisfaction across 100+ financial advisors), and the Overseas AI Agent Platform (3-layer Agent architecture, deployed in Vietnam). I hold 6 utility patents, all in LLM application system architecture. My focus is production, not demo: regulated environments, compliance constraints, and real users in real financial institutions.
Principles
The Bitter Lesson
Scale and compute win. I build systems that leverage foundation models rather than fight them.
Platform-First
One-off solutions don't survive. I design for reuse, governance, and team autonomy from day one.
Evaluation-Driven
If you can't measure it, you can't ship it. Rigorous evaluation is the foundation of every project I touch.
Stack & Skills
Credentials
Patents
6 granted (GenAI / Knowledge Graph / Data Query systems)
Award
1st Place — 2024 Enterprise Innovation Competition
Talks
iThome Hello World 2024 & 2025 (7 sessions), DevFest Taipei 2025
Certifications
PMP (Project Management Professional)
Education
NTU CS (MSc) · Quantic MBA · TU Chemnitz MSc · NTU BSc
Languages
Chinese (native), English, German, Japanese (basic)
If you're thinking about how to genuinely deploy AI Agents in a regulated financial environment — or you're building a team that needs this kind of experience — I'm open to a conversation. Whether that's advisory work, a speaking invitation, or a longer-term opportunity.
Get in Touch
Whether you need a strategy session, a prototype, or a workshop — I'd love to hear about your challenge.