NeuroShine Technology Co., Ltd.

Advanced packaging EDA tool

The Next-Gen AI-Powered EDA Integration Platform

We don't replace your EDA stack — we unify it.

AI-powered workflow automation for CoWoS, FOPLP, and 2.5D/3D IC design.

  • 600× Faster simulation
  • 69.89% Yield improvement
  • R² 0.983 Prediction accuracy
  • 10+ EDA tools connected

Interactive package teardown

Open the package, layer by layer

Microchannel Thermal CFD

Lift the lid and see where the die runs hot. InPack.AI runs advanced fluid-dynamics simulation for extreme heat management in HPC packages, and predicts the temperature field in milliseconds with AI inference.

AI-Powered Warpage Simulation

As temperature changes, substrate, interposer and dies bend because their expansion coefficients differ. InPack.AI replaces slow FEM with an ML engine that predicts warpage and stress in milliseconds, and tunes parameters with DQN for a 69.89% yield improvement.

AI-Driven Routing & 3D Integration

Pull the HBM stack apart to see through-silicon vias and micro-bumps. InPack.AI generates RDL and 3DIC paths automatically to accelerate complex packaging layouts.

Product demo

InPack AI Agent

The semiconductor industry uses 10+ specialized EDA tools, each requiring months to master. InPack.AI is our answer — a unified AI platform that orchestrates your entire design workflow, from substrate to system, in one interface.

Prof. Chi-Hua Yu, Co-founder & Head of LAiMM Lab

AI-powered multi-physics analysis

Thermal, mechanical, electrical. One view.

From substrate to CoWoS package, InPack.AI unifies all three simulations on one platform.

Our features

Modular AI-Powered EDA

We design, develop, and implement InPack.AI: an EDA platform that helps IC design teams work smarter, not harder.

Warpage analysis flow

  • Simulation Configuration: Define package structure, boundary conditions and solver targets once, then reuse them.
  • Geometry & Material Property: Parameterize geometry and materials to spin up CoWoS / FOPLP design variants fast.
  • Data Generation & Processing: Auto-generate and clean training datasets to give AI models a reliable foundation.
  • AI Model & Performance: Replace slow FEM with an ML engine that predicts warpage and stress in milliseconds.
  • DQN Reinforcement Learning: Keep optimizing parameters with deep reinforcement learning to converge on the best yield.

AI-Powered Multi-Physics Analysis

Go beyond single-point warpage prediction. InPack.AI integrates thermal, mechanical, and electrical simulations in one unified platform — covering the entire design flow from substrate to CoWoS packaging. Achieve 600x acceleration with our ML-driven engine and 69.89% yield improvement through DQN-based parameter optimization.

AI Co-Design Studio

NeuroShine AI doesn't just predict; it creates. Through a Deep Reinforcement Learning (DQN) framework, it accurately simulates multi-physics effects without a massive database, providing reliable scientific validation for your designs with an R² score of up to 0.983.

Core competencies

The Six Pillars of InPackAI Flow Suite

InPackAI Flow Suite: empowered by six technological pillars

  • AI-Powered Inference: Millisecond-level warpage & stress prediction, replacing time-consuming FEM simulations.
  • AI-Driven Routing: Automated RDL and 3DIC path generation to accelerate complex packaging layouts.
  • Workflow Orchestration: Modular "Lego-style" architecture to seamlessly connect fragmented EDA stages.
  • EDA Tool Bridge: Seamless data conversion and integration across Synopsys, Cadence, and Ansys.
  • AI Co-pilot: Natural language task dispatching powered by an advanced Skill + Memory framework.
  • Microchannel Thermal CFD: Advanced fluid dynamics simulation for extreme heat management in HPC designs.

Zero disruption

Seamless EDA Tool Integration

InPack.AI works with your existing EDA stack — not against it. Connect Synopsys, Cadence, Ansys, and Zuken tools through our unified API layer. Deploy as cloud SaaS for instant access, or on-premise for maximum data security. Either way, your existing workflow stays intact while gaining AI superpowers.

Academic & industry collaboration

Trusted by Academia & Industry Leaders

Born from LAiMM Lab at National Cheng Kung University, InPack.AI has been validated in real-world production with industry leaders including ASE and partners in the CoWoS/advanced packaging ecosystem. Our technology bridges cutting-edge AI research with practical semiconductor manufacturing needs.

Our process

Our Simple, Smart, and Scalable EDA Process

We design, develop, and implement InPack.AI to help IC design teams work smarter, not harder.

Step 1 Connect Your Tools

Integrate with your existing Synopsys, Cadence, and Ansys environments. InPack.AI acts as a unified orchestration layer.

Step 2 Define Your Workflow

Drag-and-drop modules in our AI Co-Design canvas. Or simply tell our AI assistant what you need:

Step 3 AI Executes & Optimizes

Our Multi-Agent system calls the right tools automatically.

Step 4 Get AI-Generated Reports

Receive comprehensive analysis reports, auto-generated by AI. Export to PDF, share with your team, iterate quickly.

Benefits

Why Engineering Teams Choose InPack.AI

Discover how InPack.AI enhances efficiency, reduces costs, and drives design simulation with smarter, faster processes.

  • End Tool Fragmentation: Before: switch between 10+ tools and 10+ file formats. Now: one unified platform with seamless data flow.
  • Multi-Physics Integration: Seamlessly integrates multi-physics analysis on one platform, eliminating tool fragmentation.
  • 24/7 Availability: AI-powered systems operate around the clock, ensuring seamless support and execution without downtime.
  • Cost Reduction: AI automates routing and parameter settings, reducing manual intervention and optimizing resource allocation.
  • Data-Driven Insights: Leverage AI to analyze large datasets, identify trends, and make smarter, faster, and more accurate design decisions.
  • Predictive Accuracy: The InPack.AI core engine performs instant, accurate predictions in warpage analysis, providing a reliable basis for your decisions.

About us

Who We Are

NeuroShine was founded with a vision:to build Taiwan's answer to the fragmented EDA landscape.

  • Chi-Hua Yu (游濟華): An Associate Professor at NCKU specializing in AI bionics and multi-physics simulation technologies.
  • Shin-Ruei Lin (林欣瑞): A senior software development expert from TSMC, specializing in numerical simulation methods and industrial software design.

Events & news

NEUROSHINE NEWS

FAQ

Frequently Asked Questions

How does InPack.AI EDA handle multi-physics simulation?

InPack.AI integrates thermal, mechanical and electrical simulation on one platform, replaces slow FEM solving with an ML engine, and keeps optimizing parameters with DQN reinforcement learning — giving consistent multi-physics analysis from substrate to CoWoS package.

Is our design data secure?

You can deploy on-premise so design data never leaves your environment, for maximum data security. Cloud SaaS is available when you want instant access.

Can InPack.AI integrate with our existing EDA workflow?

Yes. InPack.AI works with Synopsys, Cadence, Ansys and Zuken tools through a unified API layer, acting as an orchestration layer over your existing flow — no need to replace your toolchain.

Besides warpage, what other specific problems can your tool solve?

Beyond warpage and stress prediction, InPack.AI covers automated RDL and 3DIC routing, microchannel thermal CFD analysis, parameter optimization and AI-generated reports.

What kind of support can we expect when using InPack.AI?

We provide onboarding, workflow setup and technical support. Email syuanku@neuroshine.co to discuss your needs and we will be in touch.

Contact

Get in Touch with Us

Have questions or need EDA AI solutions? Let us know, and we'll be in touch!

syuanku@neuroshine.co

NeuroShine Technology Co., Ltd.
Room A218, 2F, No. 5, Guiren 15th Road, Guiren District, Tainan City 711010
AI Innovation Application Building, National Science and Technology Council