Wanjin Li
EN中文DE

Wanjin Li

I turn machine-learning, optimization and agentic-AI ideas into dependable software — from industrial signal tools to constrained decision systems and local-first knowledge products.

Professional focus
Data Scientist · Machine Learning Engineer · Applied AI Engineer
Explore selected workMy working principles
E-MailGitHubLinkedIn

One practice, several connected systems.

Wanjin / Partial-Discharge Detection & Segmentation

Two related but distinct tasks: classify short segments as partial discharge or noise, then segment long records into intervals of different discharge types and noise.

Selected systems / 01

Selected work, built around real problems.

Through these nine case studies, I show how I approach industrial signal analysis, energy decisions, optimization and AI systems.

01

From waveform to inspectable intervals

Partial-Discharge Detection & Segmentation

Two related but distinct tasks: classify short segments as partial discharge or noise, then segment long records into intervals of different discharge types and noise.
  • Python
  • PyTorch
  • FCN
  • CNN-BiLSTM
  • U-Time
  • Wavelets
02

Day-ahead price forecasting × constrained optimization

Energy-Storage Revenue Optimization: From Price Forecast to Daily Schedule

The system estimates the next day's 96 price points from market and weather forecasts, then selects charge and discharge windows under fixed-duration and charge-before-discharge rules.
  • Python
  • scikit-learn
  • XGBoost
  • pandas
  • NetCDF
  • YAML
03

Expert review at industrial scale

Long Time-Series Annotation Workbench

A desktop workbench for navigating, annotating and resuming work on very long measurement records while keeping model suggestions separate from expert-approved labels.
  • Python
  • PyQt
  • PyQtGraph
  • NumPy
  • Polars
  • Parquet
04

Constrained sequential decisions

Bridge Span Planning

A shared bridge-layout model with interpretable cost terms, classical baselines and masked reinforcement-learning solvers for varying bridge lengths.
  • Python
  • PyTorch
  • Beam-DP
  • DQN
  • Action Masking
  • Slurm
05

Local-first active learning

Vault Coach

An Obsidian learning companion that connects source-grounded retrieval, practice, mastery feedback and a reviewable concept-evidence graph.
  • TypeScript
  • Obsidian API
  • RAG
  • Knowledge Graphs
  • WASM
  • Vitest
06

Organizing unfamiliar signal families

Partial-Discharge Cluster Analysis

A multi-domain feature and clustering workflow that groups similar discharge candidates, preserves outliers and returns every cluster to representative waveforms for expert review.
  • Python
  • NumPy
  • SciPy
  • scikit-learn
  • HDBSCAN
  • PCA
07

Sparse data to inspectable trade-offs

Offshore Wind Design Platform

A research platform that connects sparse design data, OpenSees evaluation, evolutionary search and two-space Pareto analysis for offshore jacket concepts.
  • Python
  • Flask
  • Socket.IO
  • OpenSeesPy
  • pymoo
  • NSGA-II
08

Requirements, prototype and evaluation

Enterprise Manual RAG Thesis

A Design Science thesis on turning technical-manual questions into a source-grounded local RAG prototype and a separated retrieval/answer evaluation plan.
  • Python
  • Local Embeddings
  • Hybrid Retrieval
  • Reranking
  • LLM
  • RAG Evaluation
09

Reliable workflow before autonomous agent

AI Career Workflow Platform

A typed monorepo for structured job data, resume ingestion and observable background work, with explicit security and authorization boundaries.
  • Next.js
  • tRPC
  • Zod
  • Prisma
  • BullMQ
  • Redis
Explore all project case studies

Engineering areas / 02

From data to predictions, decisions and better data.

Data: I turn raw signals, domain context and interaction traces into inspectable inputs. Applied machine learning / deep learning: I turn high-frequency time series and business data into predictions for data-driven operations and decisions. Operations research & reinforcement learning: I move beyond prediction by searching for feasible actions under objectives, constraints and uncertainty. Knowledge bases & agent systems: I use grounded knowledge, structured outputs and reviewable workflows to support human decisions and improve the data that feeds modeling. Full-stack engineering: I connect data mining, modeling, optimization and complex agent products into usable systems.

Experience / 03

Research questions carried through to working software.

My experience moves between industrial signal analysis, data-driven structural optimization and enterprise data workflows. I implemented classification, segmentation and anomaly-detection paths; the classification interface was deployed to production, while segmentation and anomaly detection remained in the test environment.

View full experience
  1. 01Desktop high-frequency time-series annotation tool
  2. 02Records up to 560M samples · 70% profiling improvement
  3. 03ERP data flows & integration testing
  4. 04Master’s-level Business Informatics degree from a German University of Excellence

Writing / archive

Technical ideas should be teachable.

I write in Chinese about reinforcement learning, machine learning, AI agents and the engineering decisions behind working systems, while making AI understandable to a wider audience.

  • Reinforcement Learning
  • Machine Learning
  • Agentic AI
  • Engineering
Explore the editorial archive

Open to opportunities

Let’s build intelligent systems that hold up in practice.

I am interested in applied AI, machine-learning and software-engineering roles where technical depth, product thinking and careful evaluation belong together.

LinkedInGitHubE-Mail
Based in Dresden · Available in German, English and Mandarin