HANDS-ON ENTERPRISE IT & SYSTEMS INTEGRATION

Hugo Lu呂芳文 ／ 雨果

Hands-on Enterprise IT Lead · IT Specialist · Systems Integration & AI-Assisted Engineering
"Responsibility and attitude always come before expertise. If you are not afraid of hard work, there is always a chance to make something of yourself."

Nearly 18 years in enterprise IT. Currently the sole IT contact at a manufacturer of about 100 employees, attached to the Chairman's Office and reporting directly to the Chairman, coordinating IT operations, cross-department requests and vendors. From new-plant IT infrastructure and ERP integration to approval-workflow and document-management systems, I turn business needs into deliverables that people can use and maintain. I use multiple AI agents and local LLMs to assist development and troubleshooting, with an emphasis on backups, change records and real verification. "Technical lead" here describes scope of responsibility, not a formal management title.

Integrity · Diligence · Accountability Multi-agent AI orchestration across machines Primary/standby shared-memory service In-house approval & document-management systems Local multilingual LLM customer service Proxmox VE + Ceph 10G cluster New-plant relocation infrastructure
Explore the AI collaboration setup → Browse de-identified projects Contact
18 yrs
Enterprise IT & ERP experience
Manufacturing, real estate, systems integration and consulting
New plant
Infrastructure & vendor coordination
Server room and network relocation, extra network drops, acceptance
Workflow integration
Approvals, documents and ERP links
Requests, approvals, document lookup and notifications in one system
Local AI
Knowledge search & multilingual support
Answers refined through tested scope and human feedback
CORE DIFFERENTIATOR // AGENTIC AI WORKFLOW

AI-assisted engineering: treating AI as a "manageable engineering team"

Not a one-off chat window, but an enterprise-style setup with shared memory across machines, task division across models, two-way notifications, change guardrails and engineering records.

⚡ 5 machines × 4 AI agents × primary/standby memory service × Telegram command channel

An office laptop, a main workstation, a macOS edge node, a security-audit host and a home standby host are linked by full-mesh SSH key access and a self-built memory hub, so Claude Code, OpenAI Codex, Google Gemini and a local Ollama LLM can work together.

STEP 01 // RECALL & GATE

Start-of-work recall and three gates

Before work starts, relevant project context is recalled, and context size and quota are checked. If memory is out of sync or recall fails, the records are re-read and the gaps confirmed.

STEP 02 // ORCHESTRATION

Heterogeneous models working in parallel

Tasks are assigned by nature: Claude for architecture and front-end polish, Codex for deep logic and automation, Gemini for whole-system audits and cross-machine investigation.

STEP 03 // NATIVE TG BRIDGE

Native two-way Telegram bridge

On terminals and bots that have been configured and verified, replies can be pushed out and phone commands can reach an existing conversation. Delivery and continuation still depend on login, network and CLI state.

STEP 04 // DUAL CHECKPOINT

Dual-write at close and weekly audit

At the end of a session, a local JSONL and a central event record are kept for continuing the work and for weekly summaries. Dual writes and sync results must be checked; the standby mechanism is not treated as a guarantee against data loss.

✔ Verified paths support continuing across machines and AIs; login, quota and network must still be confirmed ｜ ✔ Shared records reduce information silos; unsynced content is filled in by hand ｜ ✔ Verified scope is checked against real results and logs; unverified items are labelled separately

🛡️ Engineering discipline and guardrails

  • Back up before changing, version every change: a timestamped backup is made before any code or configuration change, with a version bump, a change list and a rollback method.
  • No secret leakage: passwords, tokens and private keys are never printed in conversations, reports or notifications.
  • Standardized scripts and environments: cross-platform encoding and syntax rules are enforced, and lessons learned go into the persistent rules so mistakes are not repeated.

🔍 Honest verification and auditability

  • No "looks like it works": the AI must supply real evidence such as build results, HTTP status codes, database queries or system logs.
  • Immediate correction: when an AI made up test data or showed a cognitive bias (for example misjudging an industry), it was corrected at once and the lesson written to core memory.
  • Records are separate from deliverables: logs and weekly reports are built from retained conversations, changes and test results; the volume of records or parallel hours does not equal business value.

🧠 Local AI in practice

  • Local inference: Ollama runs on an Apple Silicon node to reduce reliance on external inference APIs. Hardware, power and maintenance still cost money, and data flow must be checked against the real configuration.
  • Enterprise knowledge retrieval (RAG): product specifications and service manuals, stress-tested with more than 5,100 questions, answer automatically in 6 languages.
  • Human-in-the-loop improvement: a technician correction-feedback platform and a rule-based terminology engine (154 term pairs) keep improving accuracy.
FLAGSHIP CASE STUDIES // DE-IDENTIFIED

Three flagship projects in depth

From relocating a server room and network to a new plant, to building the core approval and document-management systems in-house, to deploying a local multilingual AI service desk: full-stack coordination and execution across hardware and software.

🔒 Security note: all projects below are strictly de-identified. Company names, internal system codes, IP addresses, host names and personal details have been removed.
Infrastructure & project coordination

New-plant relocation: multi-vendor coordination and base IT infrastructure (extended to a second site)

As the only IT contact, in my first month I took over planning, procurement, acceptance and relocation of the new plant's network, server room, IP phone system, attendance and access control, and meeting-room AV. The server room and network were cut over on moving day, and extra network drops were added and faults fixed afterwards. This does not claim zero downtime throughout.

6+ typesof external vendors coordinated
Dozensof purchase / quote / acceptance documents
~200 hlogged on server room and network
Doneserver room and network cut over on moving day
Stage 1｜Site surveySurveyed several buildings and floors; evaluated backbone fibre, wired drops and plant-wide wireless
Stage 2｜Quotes and decisionPresented "standard" and "lean" options to management; completed negotiation and scheduling
Stage 3｜Equipment setupRacks, UPS, firewall and switches installed; cross-site VPN and failover verified
Stage 4｜Moving-day cutoverServer room moved and rebuilt; each department's network, ERP and phones brought online the same day
Stage 5｜Reinforcement and acceptanceAdded Wi-Fi coverage for blind spots and new drops; removed a LAN loop storm; completed acceptance
Stage 6｜ExtensionConsolidated and recycled old-plant equipment; repeated the approach for the second site's network

① Vendor coordination and procurement governance

  • Coordinated 6+ types of external vendors: telecom carriers, system integrators, network-equipment resellers, UPS/power and low-voltage contractors.
  • Drew up "standard" and "lean" budget options; handled comparison, negotiation, payment terms and vendor registration.
  • Followed internal procurement controls for requisition, sign-off, ordering and on-site acceptance, with all paperwork retained for audit.

② Server room and plant-wide network

  • Core server room: planned rack cabling, UPS and power improvements, and consolidated old equipment into a dedicated room.
  • Network backbone: managed switches, multi-floor wired drops, seamless plant-wide Wi-Fi, multi-WAN load balancing and cross-site VPN.
  • Peripherals: integrated the IP phone system, cross-site time clocks, large meeting-room projection and AV, and network printers.

③ Incident response and hands-on troubleshooting

  • Cut fibre during roadworks: when the old plant's telecom fibre was dug up and phones went down, a 4G backup route was activated the same day while repairs were arranged.
  • Network loop storm: traced an intermittent LAN outage to a physical loop and added protection on managed switches.
  • Recovering old assets: located servers and disks left at the old plant, inventoried historical finance data and rebuilt usable parts into standby nodes.
Full-stack system development

Internal administration, BPM e-approval and document-management platform (in-house web system)

Developed and modernized independently with ASP.NET (.NET 4.8 / C#) and MS SQL Server, closely integrated with the underlying ERP database, and iterated frequently through more than 200 production releases.

v26.200+production releases to date
6,400+documents indexed in the document centre
53 fieldswritten back to the ERP attendance table
Dual channelEmail + LINE instant notifications

① Unified approval engine (SignCore) with two-way feedback

  • One approval engine for leave, overtime, missed clock-in, computer-account access requests (IA) and hardware/software requests (ID).
  • Three-state routing (pending / not yet your turn / signed), applicant self-edit before the second step, rejection that ends the whole request, and A4 form printing.
  • A modern centred frosted-glass HUD overlay and in-place highlighted status banners replace browser pop-ups and page jumps.

② Document centre and smart meeting rooms

  • Document centre: automatically syncs more than 6,400 ISO policy documents and contracts from the file server, with online preview via Office-to-PDF conversion and contract-expiry warnings (not due / due within 3 months / expired / no end date).
  • Room booking: week-calendar view, one-off and recurring bookings (daily / weekly), attendees filled in by department, automatic sync to personal worklogs, and privacy protection for meeting content.

③ Worklog, IT queue and notification hub

  • Worklog and department view: colour-coded calendar categories, project time attribution, department worklogs for managers, and automatic write-back to IT staff worklogs when an IT request is completed.
  • Plant-wide mail and LINE webhook: an approval-email queue with a send-audit page, plus integration with the company's official LINE account webhook for instant push notifications.
Enterprise AI in production

Local multilingual AI service desk and knowledge base

A local Ollama large language model runs on an Apple Silicon (M4) edge node, with a Python RAG retrieval back end and a front-end I built, integrated with the company's systems through an HTTPS reverse proxy and single sign-on.

$0cloud API token fees
5,134 questionsverified by automated test reports
6 languagesChinese, English, Japanese, Vietnamese, Polish, Spanish
154 pairsforced industrial-term corrections (Traditional/Simplified)

① Local LLM and enterprise knowledge retrieval

  • Knowledge retrieval and Q&A run on local inference, reducing dependence on external inference. Hardware, power, maintenance and external integration still count as cost, and data flow is checked against the deployment settings.
  • Batch stress tests and quality audits over more than 5,100 questions; fixed truncated long answers and mixed Chinese-English output.

② Language detection and terminology alignment

  • Automatic detection of Traditional and Simplified Chinese, English, Japanese, Vietnamese, Polish and Mexican Spanish, with replies in the matching language.
  • A built-in rule-based terminology engine (154 term pairs on record) improves Traditional/Simplified output and technical wording. Testing and human review are still needed; zero error is not guaranteed.

③ Technician feedback platform and SSO access

  • Built an "AI service log and technician correction platform" with multi-recipient notifications, back-filling of missed entries and version history (nearly 50 iterations).
  • Integrated with the administration system through an HTTPS reverse proxy and role-based access control (RBAC) for secure internal and external access.
ENTERPRISE PORTFOLIO // 12 KEY PROJECTS

Enterprise project overview (de-identified)

Virtualization clusters, cross-site SD-WAN, security and encryption governance, 3-2-1 off-site backup, attendance and CAD engineering automation, and more. Click a category to filter:

Virtualization and high-availability storage✓ Completed · in maintenance

Multi-node Proxmox VE virtualization cluster with Ceph distributed storage

Built a three-node Proxmox VE cluster upgraded to 10G fibre, with Ceph distributed storage, consolidating old physical servers and mixed virtual machines.

  • Live-migrated the core ERP host, SQL database, web host and two domain controllers (DC) without interruption.
  • Resolved a cluster file-sharing deadlock and memory bottlenecks; set up daily snapshots and a backup-space monitoring SOP (over 230 hours in total).
Proxmox VE Ceph Storage 10GbE fibre Hyper-V migration Linux HA
Network architecture and deep troubleshooting✓ Completed

Layered firewalls, cross-site SD-WAN and routing-hijack root-cause analysis

Designed layered firewalls, multi-WAN failover and cross-site SD-WAN tunnels, and resolved several hidden network outages that had troubled the company for a long time.

  • Found the root cause of repeated domain-trust failures and unreachable same-subnet devices: virtual VPN software hijacked DNS with NRPT rules and its subnet route metric overrode the physical NIC.
  • Traced a cross-subnet outage to a viewer machine given a wrong static IP, which triggered Windows RFC 5227 address-conflict protection, and fixed it remotely through system services without disruption.
Layered firewalls SD-WAN / VPN AD / DNS / DHCP Route metric tuning
Security hardening and compliance governance✓ Completed · ongoing

Enterprise security: DLP drawing encryption, AD red-team assessment and SIEM monitoring

Defence in depth, from endpoint encryption and domain privilege governance to centralized log monitoring.

  • Led the rollout of engineering-drawing encryption software (over 150 hours of intensive deployment) and resolved CAD and office-software compatibility issues.
  • Ran a domain controller (AD) vulnerability assessment, trimmed privileged accounts and disabled legacy protocols; pushed mandatory SSH-key login across the fleet and NAS.
  • Planned capacity and automated deployment for a 70-node Wazuh SIEM platform, and set up tracing for confidential printing.
DLP document encryption AD red-team assessment Wazuh SIEM SSH key governance
Backup system and off-site recovery✓ Completed · expanding

3-2-1 backup redesign, cross-site off-site backup and "silent failure" root-cause fix

Audited and rebuilt the company's backup system, closing a serious blind spot where the interface showed jobs as enabled but nothing actually ran.

  • Found that on two main NAS units the schedules looked enabled, but cron registration failures and a firmware Python environment defect had silently stopped them for over 3 months. Fixed the cause and added a verification mechanism.
  • Built a cross-site off-site backup node and a full company mailbox backup, and wrote an hourly RAID5 rebuild monitor that reports sync progress (%).
Off-site NAS backup Rsync / Snapshot RAID monitoring Mailbox backup
Process automation✓ Live (95%+)

End-to-end attendance automation and live status board across plants

Built a fully automatic pipeline from physical time clocks to a read-only collector, data cleaning and the ERP attendance table, replacing daily manual imports.

  • Designed dynamic scheduling: every 5 minutes at shift peaks (07:45–08:15) and every 30 minutes otherwise; offline clocks are skipped without blocking.
  • Idempotent 15-minute writes to the ERP prevent duplicates, with automatic back-fill after a disconnection, and a built-in "clock collection log and online status" monitoring page in the management system.
Python MS SQL / ERP integration Idempotent scheduling Live monitoring board
Engineering automation✓ Live (90%+)

Batch 3D/2D drawing (CAD / Inventor) to PDF automation

Removed the time-consuming, error-prone manual conversion of large numbers of design drawings with a reliable batch output pipeline.

  • Found in testing that the same file extension does not mean the same internal format, so files are routed to the right engine by reading the binary header.
  • Worked around the limits on CAD COM components in a Windows non-interactive session (Session 0); in that test sample, 75 of 75 complex drawings were output. This does not mean every drawing behaves the same way.
Autodesk Inventor CAD batch conversion Binary header parsing Web drawing selection
Core system development✓ 200+ iterations

Internal administration and BPM e-approval system (full-module modernization)

Independently maintain and extend the company's core administration portal, covering HR attendance, e-approval, document centre, meeting rooms and the IT service desk.

  • Fixed dozens of deep defects in the legacy system, including a session-validation hole, attachments crossing between form types, and iPhone EXIF photo rotation and cropping.
  • Implemented A4 standard printing for all forms, a batch permission matrix, plant-wide email notification switches and instant notifications through the official LINE account.
ASP.NET (.NET 4.8) C# MS SQL Server JavaScript / IIS
Drawing management and system evaluation✓ Completed

PDM/PLM acceptance review and Autodesk Vault deployment

Led project acceptance, alternative-solution feasibility and access governance for the R&D design department's drawing-management system.

  • Completed the acceptance report for the existing PDM/CAD project, root-cause analysis of packaging failures, and cross-system BOM data integration.
  • Independently deployed Autodesk Vault Server with a go-live guide, and set up tiered read-only access to 3D assembly files on the NAS.
PDM / PLM evaluation Autodesk Vault BOM data integration ACL governance
AI infrastructure✓ Built and maintained

Multi-agent AI collaboration network with a primary/standby high-availability memory service

A self-built collaboration network across 5 hosts and 4 kinds of AI agents, with memory synchronized across devices and live two-way control from a phone.

  • Built primary and standby memory servers (automatic failover after a 2-second timeout), supporting semantic and keyword recall and end-of-session event sync.
  • Built a native terminal Telegram bridge that continues existing conversations. Login, quota, network and delivery still have to be checked, and parallel records are not converted into equivalent headcount.
Claude / Codex / Gemini Dual-Brain HA Telegram Bridge Tailscale / SSH
Local AI application development✓ Live (90%+)

Local multilingual AI service desk and bilingual product-catalogue engineering

Combines a local LLM service desk with multilingual product-specification pages to strengthen international sales and after-sales service.

  • Deployed a local AI service desk with multilingual Q&A, technical-term correction and technician feedback. Language and question-bank coverage follow the existing test records.
  • Helped build an interactive Chinese/English two-tab module for the main machines' features and specifications, embedded directly in the website and sales process.
Ollama Local LLM Python RAG HTTPS reverse proxy Multilingual engineering
Digital forensics and data recovery✓ Completed

Sector-level (MFT / USN / PST) digital forensics and critical data recovery

For damaged disks, hosts that rebooted unexpectedly and past security incidents, wrote low-level parsing tools to carry out read-only forensics and recovery.

  • Parsed NTFS MFT, USN Journal, Windows Event Logs (EVTX) and a corrupted large Outlook PST file directly in read-only mode, recovering key HR and business mail.
  • Produced a sector-comparison and timeline forensic report on a past ransomware encryption incident to strengthen the company's defence baseline.
NTFS MFT / USN forensics PST low-level recovery EVTX log analysis Python / C#
Data analysis and policy governance✓ Completed

ERP raw-material cost and purchasing analysis; standardizing company processes and work rules

Supported finance, purchasing, international sales and administration, turning complex data and policies into decisions that can be acted on.

  • Found why ERP raw-material costs would not update, then produced reports on lead times, latest purchase prices and price variance for tens of thousands of items.
  • Used international LEI data and multilingual search to compile prospective customer lists in Europe and the Americas (Poland, Italy, Mexico and others) and a cross-border meeting phrasebook; completed a review of the company work rules (72+ hours) and documentation of business processes (60+ hours).
ERP cost data analysis SQL data extraction Work rules & SOPs Cross-border sales data
QUANTITATIVE METRICS // WORKLOG ANALYTICS

Engineering records and how they are counted (Apr 2025 – Nov 2026)

Compiled from the existing worklogs, which originally listed 2,223 entries and 3,789.4 hours, to show how time was spread across operations, infrastructure and development. These are engineering records, not cost-benefit figures or independent person-hours: parallel AI work may overlap, the data has not been de-duplicated or re-audited, and it is not used to estimate hours saved, equivalent headcount or ROI. Release counts, project counts, document volume and test-question counts describe scale only, not results.

📊 Logged hours by area (by work category, not de-duplicated)

Frequent day-to-day user support (over 1,000 entries) alongside deep system development, virtualization and security projects
IT accounts / endpoint support
1,371 h
Virtualization / server infrastructure
302 h
Software evaluation / rollout
292 h
Systems development (admin / approvals / automation)
285 h
Backup / disaster recovery / data rescue
275 h
Equipment faults / repair
248 h
Security hardening / encryption rollout
237 h
Network / server room / relocation
197 h
AI collaboration / local LLM / memory service
148 h
Cross-department meetings / training / policy
136 h

📈 Monthly logged hours (including parallel AI records)

Peak month in the original statistics: 329 logged hours. Includes parallel overlap; not independent person-hours or equivalent engineering output
36
140
109
149
282
194
217
162
226
146
242
178
267
236
280
329
310
25/4568910111226/1234568910

💡 From "one person does everything" to "one person leading a team of AI engineers": routine scripting, log screening, report generation and front-end layout are handed to several AI agents working in parallel, while I focus on architecture decisions, vendor communication, on-site hardware troubleshooting and final acceptance, well beyond what one person's time would allow.

CAREER & COMPETENCIES // 18 YEARS

Career history and technical skills

A business-administration degree combined with nearly 18 years of front-line IT work, from management support and cross-department coordination to putting systems into production. The periods below follow the existing resume. Consulting / self-employed work overlaps with other jobs; years are counted along the timeline and not added twice, and no part-time or full-time status is implied.

Apr 2025 – present

IT Specialist (enterprise IT coordination and hands-on work)

Manufacturer of about 100 employees｜Taoyuan, Taiwan｜Chairman's Office, reporting directly to the Chairman
  • Sole IT contact, coordinating cross-department needs, outside vendors and the multi-site IT environment: Proxmox/Ceph virtualization cluster, layered firewalls and SD-WAN, 3-2-1 off-site backup, AD domain and security encryption, endpoint support and in-house system development.
  • In my first month, led the new-plant IT infrastructure and vendor coordination, completing the server room move and cross-site network go-live on schedule, and then extended the work to a second site.
  • Independently developed and keep iterating the administration and e-approval system (200+ releases), a local multilingual AI service desk, time-clock attendance automation and a batch CAD drawing output system.
  • Built a multi-agent AI collaboration setup with a primary/standby memory service, and manage 25+ projects with project-based tracking and worklogs.
Dec 2017 – Sep 2024 (6 years 10 months)

Management Staff / ERP Developer (Software Engineer)

Asset management and commercial real estate company｜management and staff functions
  • Planned and introduced management policies, information systems and standard operating procedures (SOPs); independently developed internal business and management systems.
  • Led digital marketing, data analysis and SEO keyword optimization to support management decisions; planned recruitment processes and sales training.
  • Responsible for servers, network equipment and firewalls, as well as the company website and Google Business listings.
Jan 2014 – Dec 2024 (11 years)

Business Consultant / Operations Manager, Systems Integrator, ERP Project Specialist

HUGo Studio (hugoi.com)｜computer systems integration and consulting (overlaps with other jobs)
  • Provided ERP rollout, custom development and digital-transformation consulting to small and medium-sized businesses; planned IT management policies and project collaboration.
  • Led corporate-image and e-commerce website planning, online marketing and social media, as well as hosting, servers and corporate firewalls.
Oct 2008 – Oct 2014 (6 years 1 month)

Project Manager / Operations Manager, Systems Integrator, ERP Project Specialist

IT and systems integration company｜managed a team of 5 to 8
  • Managed the administrative and technical departments and handled cross-department decision-making and coordination; developed enterprise systems, designed management policies and drove large customer projects to delivery.
  • Responsible for presenting and selling commercial software products, allocating department sales targets and meeting project performance goals.

🤖 AI engineering and cross-machine automation

Claude CodeOpenAI CodexGoogle Gemini / AGY Ollama local LLMRAG knowledge retrievalPrompt Engineering Primary/standby memory service (Dual-Brain HA)Telegram bot two-way bridge

💻 Software development and databases

ASP.NET (.NET 4.8)C#Python JavaScript / HTML5 / CSS3PowerShell / Bash MS SQL ServerMySQL / MariaDBIIS / Nginx GitRESTful API / Webhook

🖧 Virtualization, storage, network and security

Proxmox VE HA clusterCeph distributed storageVMware / Hyper-V Windows Server / Active DirectoryLinux (CentOS / Debian / Kali) macOS (Apple Silicon)Layered firewalls / IPSSD-WAN / VPN 10GbE fibre networkingNAS / SAN off-site backupWazuh SIEMDLP document encryption

🏢 Enterprise ERP and engineering applications

Lingyue ERP (凌越)Wenzhong ERP (文中)Digiwin ERP (鼎新)Qianao ERP (千奧) BPM e-approval designAutodesk Vault / Inventor automation PDM / PLM evaluationDigital forensics and data recovery (MFT/PST)

🎓 Education and certifications

• Kainan University, Department of Business Administration (bachelor's degree)
• Qiying (啟英) Senior High School, Data Processing program (high-school diploma)
• Taiwan Class C technician certificates in computer software design and in accounting
• Microsoft Office Specialist (MOS Excel) international certification
• TQC-OA Chinese and English typing certifications
HUGO STUDIO // CONSULTING & SOLUTIONS

HUGo Studio: core skills and consulting areas

The same conviction, year after year. Combining 18 years of enterprise experience with current AI-assisted engineering, to deliver IT solutions that fit the need, can be maintained and state clearly what has been verified. Cost and benefit are assessed case by case.

🤖

Enterprise AI collaboration and local LLM setup

Help companies evaluate a local AI service desk and internal knowledge base (Ollama + RAG) and multi-agent AI workflows, planned around data security, hardware and maintenance. Total cost is not claimed to be zero.

⚙️

ERP integration and custom internal systems

ASP.NET / Python internal administration portals, BPM e-approval, document centres, attendance automation, and two-way links to legacy ERP databases.

🛡️

HA virtualization clusters and off-site backup

Proxmox VE + Ceph HA cluster migration, 3-2-1 cross-site backup planning, coordination of new-plant server room and network relocation, and forensic recovery of damaged disks and mailboxes.

🌐

Network security and IT consulting

Layered firewalls and multi-site SD-WAN, AD security hardening, DLP drawing-encryption rollout, SEO-friendly corporate websites and long-term IT consulting and operations.

GET IN TOUCH // CONTACT HUGO

Open to talking technology, projects and collaboration

Whether it is upgrading enterprise IT architecture, managing an MIS department, building internal systems and ERP integrations, or putting multi-agent AI teams and local LLMs to work in day-to-day operations, you are welcome to reach me by phone, email or LINE.

📞 Mobile: +886 970-071-005
ContactHugo Lu (呂芳文 ／ 雨果)
Mobile+886 970-071-005
Office line+886 3-325-6688
Consulting emailinfo@hugoi.com
LINE ID0970071005
Add on LINE
LocationLuzhu District, Taoyuan, Taiwan (remote and cross-site work supported)