Skip to content

Industrial AI Infrastructure Architect

Robin Ji

纪多斌

18 years. 7 industries. One thread.

Shanghai, China GitHub YouTube
See projects Contact

About

A translator,
not a bystander

Industrial software → Cloud → AI

18+ years walking the industrial-software value chain: CAD/CAE tools vendor → PLM platform → shop-floor MES → digital thread → CAE enterprise account management → automotive full-stack digitalization → cloud + AI infrastructure.

My core skill is converting engineering problems into systems that scale — wind-turbine structural issues, battery-pack quality concerns, GPU cluster capacity — solved through architecture.

Now focused on industrial AI infrastructure: CAE-on-cloud, AI for Engineering, Agent × CAE orchestration — Ansys SimAI™, Siemens Xcelerator AI, Dassault 3DEXPERIENCE AI.

Highlights

By the numbers

Evidence over adjectives

01

44★

GitHub flagship project

`document-superpowers` — 4-stage AI writing skillset (Brainstorm → Plan → Execute → Review) for Cursor / CodeBuddy / Claude Code.

02

18+ yrs

Industrial software → cloud → AI infra

Siemens → GE → Ansys → JLR → TE → Carestream → leading Chinese cloud vendor. Hands-on, not bystander.

03

7 releases

CloudSim simulation cloud 0→1

Delivered to 20+ enterprise clients (CSG, SJTU AERI, COMAC Shanghai Aircraft Design Institute) — CFD / structural / STA / DRC / SI-PI / EMI-EMC / financial compute.

04

93 videos

YouTube NX CAD/CAM/CAE tutorials

@binbinao · 234 subscribers · ~55K views · 5-year evergreen on 5-axis machining, DMU kinematics, mold design.

Experience

Across three eras:
industrial software, cloud, and AI

  1. 2020 — Now

    Shanghai

    Leading Chinese Cloud Vendor

    Head of Computing Architecture / Simulation Cloud Product Owner

    • Led Shanghai Electric Wind Power HPC cloud migration in 2020, using it as a springboard to transition from architect to Simulation Cloud PM; validated HPC-cloud feasibility with one flagship customer.
    • Owned CloudSim simulation cloud 7-release iteration (v1.0 → v1.7, 2021–2022) across financial / education / automotive / industrial sectors; customers include CSG, COMAC Shanghai Aircraft Design Institute, and SJTU AERI.
    • Serving government & enterprise clients since 2022, led GPU-compute scenario rollouts. Customer focus is "what GPU can solve", not hardware specs. Notable cases: Bosch autonomous-driving training / ARM China EDA scheduling / Novita.ai elastic inference.
    • Full-stack AI expertise: end-to-end understanding (training / inference / application) developed through hands-on GPU deployment for gov/enterprise clients — the foundation for industrial AI and AI for Engineering.
  2. 2018 — 2020

    Shanghai

    Chery Jaguar Land Rover

    Director, Enterprise Architecture & Engineering Digitalization

    • Engineering IT Director: managed 6 architects + Engineering & Quality IT (12 total). Led engineering capability import from UK side, supporting NEV BEV launch in China — flagship models: Range Rover Evoque / Discovery Sport. Built/imported 5 systems: Teamcenter (PLM) / OTA / HPC cluster / 3-electric traceability / compliance-SAP integration — completed within 18 months.
    • Promoted to Enterprise Architecture Director: built Chery Jaguar Land Rover IT architecture blueprint covering Infra → ESB → 4 Capability Pillars (Engineering IT / Manufacturing IT / Corporate Operation / Sales & Marketing DMS); constructed the "China-side independently operable" IT capability foundation for the JV.
    • Designed TOGAF ADM enterprise-architecture blueprint across infra / data / system / app domains; gave the group a unified technical vocabulary for digital transformation.
    • Built PD-department Tableau BI cockpit: 4-dimensional visualization (project progress, cost, ROI, budget overrun); supports monthly management reviews.
  3. 2018

    Shanghai

    ANSYS (now Synopsys)

    Senior Technical Account Manager (Enterprise)

    • Owned digital-twin engagements with ABB / Bosch Automotive Electronics / Continental Automotive Electronics — 7+ Fortune-500 enterprises under mixed strategic-cooperation and standard-service models. ABB / Continental / Bosch were flagship digital-twin accounts.
    • Led ABB dry-type transformer auto-design coupling system (ANSYS product ecosystem): thermo-structural coupled simulation completed in one pass during early design; shortened product-development cycle by 50%.
    • Led Continental Automotive IGBT-box digital-twin simulation (ANSYS Icepak + CFD): well-designed Ansys tool flow enabled early-stage electrothermal simulation; design cycle shortened by 30%+.
    • Bridged industrial CAE practice with academic research at SJTU / Tongji; drove ANSYS China localization.
  4. 2016 — 2017

    Shanghai

    GE China Global Research Center

    Senior Technical PM · Digital Thread

    • Wind-power digital verification platform (highest-value deliverable): built on Ansys simulation tools; full-system simulation cycle 75 → 15 days (↓80%); won time advantage during China's wind-power installation rush; widely applied to GE wind-farm bidding.
    • Led Digital Thread strategic planning, built the "Digital Engineering → Digital Manufacturing" technology matrix; provided a unified digital foundation for GE China's coal-power and wind-power business lines.
    • Coal-power digital factory: simulation pipeline + automated reporting cut ramp-up from 45 → 15 days (↓67%).
    • VR digital radiology PoC (HTC Vive hospital radiology layout simulation), in collaboration with GE Healthcare.
    • Led 7-person digital product team (simulation engineers + data scientists + full-stack developers).
  5. 2014 — 2016

    Shanghai

    TE Connectivity

    IT Program Manager, Smart Manufacturing

    • Led 20+ member MES team; completed MES rollout across 4 workshops / 200+ machines (injection / stamping / electroplating / assembly) within 24 months.
    • Appliance ignition MES: ±5% energy monitoring, stamping OEE +35%, electroplating compliance 100%; saved ¥5M/yr in energy costs.
    • Automotive Hydra MES (18 months): SCADA second-level collection, anomaly response 15 min (vs 60+ min in legacy MES).
    • Tool migration: 2D AutoCAD → 3D NX + parametric template library; design reuse +30%; certified 50+ engineers on NX.
    • Delivered 40+ technical trainings, including 10+ advanced NX certifications.
  6. 2011 — 2014

    Shanghai

    Carestream Health

    PLM Solution Lead / R&D Effectiveness Director

    • Independently led Aras PLM (internally "PLI" — Product Level Information) deployment across China / US / France: 16 modules, 10 legacy systems retired in 3 years.
    • Innovated "Week-1 Onsite Agile Support" service: under cross-timezone / cross-cultural challenges, sat side-by-side with customers for the entire first week after go-live, avoiding friction with internal ticket-based support systems; helped customers ramp up quickly.
    • Change cycle 14 days → 72 hours (↓79%), design release efficiency +40%.
    • Saved $2M+/yr in legacy maintenance ($6M+ cumulative 3-year ROI).
  7. 2007 — 2011

    Greater China

    Siemens PLM Software

    Senior Application Engineer · NX CAD/CAM/CAE & Teamcenter

    • NX CAD/CAM/CAE + Teamcenter technical support and promotion across Greater China: 100+ enterprise customers (including Foxconn / BYD), 20+ trainings/yr.
    • Trained large customer base on NX CAD/CAM/CAE. Most memorable: a military customer in Chengdu — CAM software solved critical-part machining challenges.
    • Designed Teamcenter multi-version user-testing system; built requirement funnel connecting field feedback with global PMs; mechanism adopted by Teamcenter global product line.

Projects

From industrial software to AI agents — 5 picks

All repositories

PhysicsNeMo: From Zero to One

Hands-on tutorial book for NVIDIA PhysicsNeMo · 8 chapters + 4 appendices

8 ch + 4 app textbook size 3 variants/ch raw PyTorch / SDK / GPU 30+ commits over 3 months 2 weeks multi-agent output
  • PINN and neural-operator three-layer teaching: each chapter delivers raw PyTorch (educational) / PhysicsNeMo SDK (CPU-verified) / GPU production (DDP+AMP)
  • 6-week study path + cloud-GPU deployment guide + 50-question troubleshooting + Colab quickstart
  • Fully bilingual textbook (book/ + book_en/), LaTeX-formula friendly
  • CAE review rounds 1/2/3: 3-stage external CAE expert feedback iteration (CFD batch run / joint inversion / multi-objective optimization)
Python ≥3.10PyTorch ≥2.3NVIDIA PhysicsNeMo 2.0DDP + AMPMulti-Agent Orchestration
Tech Author / Lead Editor (personal project)
2026-05 — 2026-08 Detail

DeepDraw

DFM multi-Agent system for sheet-metal & machining

5-Agent pipeline Phase 1–7 delivery 8.8 commits/wk pace
  • Spec Interpreter → Drawing Auditor → BOM Generator → Process Recommender → Chief Verifier — 5-Agent collaboration
  • LangGraph orchestration + TypedDict state machine + self-play Reflection Loop (max 3 rounds)
  • ChromaDB enterprise-standard RAG, metadata filter for on-site deployment
  • PoC validated across 5 real scenarios
PythonLangGraphChromaDBVision-LLMPDF/DXF parsing
Full-stack (personal project)
2026-06 — 2026-07 Detail

Material-LIMS

Material lab information system · flagship industrial-software full-stack project

8 modules domain coverage 50 commits code output 10.0 commits/wk pace
  • Full workflow: commission → analysis → report → workflow → audit → equipment → calibration → SSO
  • BPMN 2.0 workflow engine replacing hardcoded state machines
  • Microsoft 365 online editing (Word) + Azure AD SSO
  • Immutable audit trail + e-signature foundation (FDA 21 CFR Part 11 compliance-oriented)
Java 17 + Spring Boot 3.2Flowable BPMN 2.0PostgreSQL 15 + Redis 7MinIOReact 18 + TypeScriptAnt Design Pro 6Azure AD + Microsoft Graph API
Full-stack / Tech Lead (personal project)
2026-06 — 2026-07 Detail

document-superpowers

4-stage document-writing skillset for AI coding tools

44★ GitHub stars 12 forks community forks 1 PR external contribution Top 5% self-built ranking
  • 4-stage workflow: Brainstorm → Plan → Execute → Review
  • Cursor / CodeBuddy / Claude Code — three-platform compatible
  • Bilingual README (Chinese + English)
  • Cross-platform filename compatibility (Windows char-handling fix from community PR)
MarkdownCursor Skill systemCodeBuddy Skill systemClaude Code Skill system

excel-ai-connector

Excel Office Add-in · multi-provider AI gateway

4 Providers model support 3 outputs modes 6 fixes/day intensive iteration
  • OpenAI-compatible / Azure OpenAI / Anthropic / Gemini unified Adapter
  • Task-pane frontend + Node.js Bridge backend (API key never exposed to browser)
  • Three output modes: overwrite selection / write new sheet / output to result box
  • Config lock prevents accidental edits to Endpoint / Model / API Key
Office.jsJavaScript / TypeScriptNode.js BridgeProvider Adapter patternlocalStorage

Skills

Five capabilities, one thread

Business Translation

01
  • Convert engineering / manufacturing / R&D pain points into executable systems — across 7 industries.
  • Full-chain driver: business communication → architecture → delivery.

Industrial Software Stack

02
  • CAE: ANSYS Mechanical / Fluent / LS-DYNA / Icepak.
  • PLM: Teamcenter / Aras.
  • MES: Hydra / Ignition.
  • CAD/CAM: Siemens NX full suite.
  • Digital Thread / Digital Twin.

AI & Acceleration

03
  • LLM training (H800, 1,000-card clusters) & inference (H20).
  • Agent orchestration (Multi-Agent / MCP / LangGraph).
  • AI for Engineering (Ansys SimAI™ 2026 hands-on).
  • AI-assisted dev: vLLM / RAGflow / LangChain.

Cloud & Infrastructure

04
  • GPU / HPC cluster design & tuning.
  • Bare-metal & virtualization.
  • CI/CD, DevOps (CloudSim v1.0 → v1.7 across 7 releases).

Delivery & Leadership

05
  • 5–20+ cross-functional teams.
  • Program management / TOGAF ADM enterprise architecture.
  • 40+ technical trainings, 10+ certified engineers.

Education

Master of Mechanical Engineering

  • Tongji University

    Master of Mechanical Engineering

    EDU

  • University of Shanghai for Science and Technology

    Bachelor of Mechanical Engineering

    EDU

Languages

Three working languages

  • Mandarin Chinese Native
  • English Professional Working (CET-6 / TOEFL)
  • German CGT-4, work document reading

Contact

Send a message

Open to conversations on collaboration, interviews, consulting.

01 · Email ••••••••••••

Obfuscated by JS — not crawlable

Public profile. Detailed project cards, STAR narratives, and compensation discussion via email.