Applied AI / Machine Learning Engineer
Time Series · Industrial Signal Processing · LLM / RAG · Optimization · AI Engineering
I build practical AI systems for industrial data analysis, knowledge retrieval, decision optimization, and intelligent engineering workflows.
I am an applied AI and machine learning engineer focused on turning algorithms into usable systems.
My work combines deep learning, time-series analysis, industrial signal processing, optimization, and LLM-based software engineering. I am especially interested in building end-to-end AI workflows: from data processing and model development to backend integration, local AI deployment, and practical user-facing tools.
Currently, I am working on projects around industrial time-series modeling, reinforcement learning for engineering design, energy storage revenue optimization, local-first RAG systems, and AI-assisted career development platforms.
- Industrial time-series analysis and signal processing
- Partial discharge detection with HFCT sensor signals
- End-to-end time-series segmentation and sequence labeling
- Reinforcement learning for engineering design automation
- Energy storage arbitrage and profit-oriented machine learning
- Local-first RAG systems, Obsidian plugins, and intelligent knowledge tools
- Full-stack AI applications with TypeScript, Python, and LLM workflows
| Area | Technologies |
|---|---|
| Programming Languages | Python, TypeScript, SQL, Java |
| Machine Learning / Deep Learning | PyTorch, scikit-learn, NumPy, Pandas, Polars |
| Time Series & Signal Processing | feature engineering, segmentation, clustering, industrial signal analysis |
| LLM / RAG | local RAG, hybrid retrieval, reranking, Ollama, OpenAI-compatible APIs, local LLM deployment |
| Backend & Web | Flask, FastAPI, REST APIs, Next.js, Node.js |
| Desktop & Plugin Development | PyQt, PyQtGraph, Obsidian Plugin Development |
| Optimization & Algorithms | Reinforcement Learning, Dynamic Programming, Genetic Algorithms, NSGA-II, PyMoo |
| Visualization | Plotly, Matplotlib |
| Engineering Tools | Git, Linux, Docker, pnpm, ESLint, Prisma |
| Project | Description | Tech |
|---|---|---|
| AI Career Agent Platform | A full-stack AI career assistant platform for job management, resume management, knowledge-base assisted matching, and interview preparation. Designed as a TypeScript monorepo and gradually evolving toward a deployable SaaS product. | TypeScript, Next.js, Node.js, Prisma, SQLite, LLM workflow |
| Vault Coach | A local-first Obsidian plugin for intelligent knowledge-base Q&A. It supports hybrid retrieval, query rewriting, reranking, incremental indexing, long-term memory, and optional OpenAI-compatible external LLM APIs. | TypeScript, Obsidian API, RAG, Ollama, local LLM |
| Energy Storage Arbitrage Optimization | A profit-oriented machine learning pipeline for energy storage revenue optimization in electricity markets. The project converts price forecasts into executable battery charge/discharge schedules and uses arbitrage profit as a key validation metric. | Python, ML, feature engineering, time-series validation, strategy optimization |
| Reinforcement Learning for Bridge Pier Layout Planning | A reinforcement learning project for generating feasible bridge pier layout strategies under different bridge lengths and restricted-area constraints. The project compares RL with rule-based, greedy, and dynamic-programming baselines. | Python, RL, DP, optimization, simulation environment |
| HFCT Signal Annotation Tool | A desktop annotation tool for labeling long industrial time-series signals and suspicious waveform segments in partial discharge inspection workflows. | Python, PyQt, PyQtGraph, time-series visualization |
| Partial Discharge Detection Model | Deep learning models for partial discharge / noise classification and waveform understanding on industrial sensor data. | PyTorch, signal processing, time-series modeling |
| End-to-End Time-Series Segmentation | A semantic segmentation pipeline for industrial time-series data, focused on sequence labeling and boundary-aware waveform recognition. | PyTorch, sequence modeling, segmentation |
| Time-Series Clustering and Feature Engineering | Machine learning workflows for clustering, representation learning, and feature extraction from industrial signal data. | scikit-learn, NumPy, Pandas, Polars |
| Structural Optimization Platform | A full-stack structural optimization system with interactive frontend-backend integration and evolutionary optimization algorithms. | Flask, PyMoo, NSGA-II, GA, Plotly |
| Enterprise Manual RAG Prototype | A local RAG prototype for enterprise product manuals with modular retrieval, reranking, and answer generation. | Python, RAG, Ollama |
- Building reliable AI systems instead of isolated model demos
- Designing reproducible ML experiments and evaluation pipelines
- Applying AI to industrial inspection, energy systems, and engineering design
- Developing local-first AI tools with privacy-aware workflows
- Combining classical optimization, reinforcement learning, and LLM-based systems
I am interested in roles related to:
- Machine Learning Engineer
- Applied AI Engineer
- Data Scientist
- Industrial AI Engineer
- AI Software Engineer
- Time-Series / Signal Processing Engineer
- LLM / RAG Application Engineer
I am particularly interested in teams working on industrial AI, intelligent software systems, time-series learning, optimization, and practical LLM applications.
GitHub: Wanjin5508
LinkedIn: wanjin-li5508
Email: liwanjin5508@gmail.com