Open-source observability for PyTorch training runs.

Generic modular factory takt-time simulator for process routes, buffers, transfer rules, bottlenecks, and capacity reports.

An open-source live telemetry and system modeling engine for generative AI. Track total system resource utilization in real time across storage, memory paths, and compute blocks to eliminate black-box execution failures and accelerate the development of next-generation AI and generative AI architectures.

Geomorphological investigation of the Wimbach and Klausbach valleys using ERT, TLS, GIS, geomorphic mapping, and Structure-from-Motion techniques to explore surface and subsurface landscape characteristics.

Measure and visualize why LLM inference is slow: bottleneck analysis, model dissection, KV-cache, GEMM/GEMV, quantization, and memory-bound decoding.

20 Hands-on Linux performance tuning labs across physical, virtual, and cloud environments covering observability, bottleneck analysis, kernel tuning, and optimization.

An advanced Python profiling and performance analysis toolkit that unifies timing, resource monitoring, line-by-line profiling, flame graphs, call graphs, memory leak detection, async/thread analysis, live dashboards, and intelligent optimization suggestions in one extensible platform.

Customer-support operations control tower: process mining, SLA bottleneck detection, and capacity forecasting on 250K+ workflow events across 50K cases. Airflow + dbt + DuckDB + Pandas + 6-page Streamlit dashboard.

Client-side GTA Online empire evaluator: income scoring, bottleneck detection, prioritized action plans, and a 795-vehicle garage database. React 19 + Vite 7, zero backend.

Interactive browser simulator for exploring how AI capability growth becomes real-economy output, or fails under physical and institutional bottlenecks.

Cinematic, high-performance HTML5 Canvas engine that visualizes database systems under load. Demonstrates failure, scaling, replication, caching, and optimization through real-time animation. Zero dependencies, production-grade rendering, and system architecture storytelling in a single-page app.

Cinematic, high-performance HTML5 Canvas engine that visualizes database systems under load. Demonstrates failure, scaling, replication, caching, and optimization through real-time animation. Zero dependencies, production-grade rendering, and system architecture storytelling in a single-page app.

Transparent, reproducible wafer-fab operations analytics: a hand-built synthetic DES validated against known ground truth, used for process mining, bottleneck proof, KPI and equipment monitoring, and CRN-paired what-if decision support.

A.F.O artifact for bridge-sensitive bottleneck attribution and control in hierarchical LLM memory paths (gem5-based reproducibility bundle).

白毛股神的供应链瓶颈分析框架 - 5-layer supply chain bottleneck investment framework

Dependency-light OEE and bottleneck analysis for manufacturing event logs.

Evidence-bound workflow diagnostics and certified lower-bound reporting for long-running AI agent pipelines. Improve agent workflow throughput without changing the model.