Systems Biology
Genome-scale and enzyme-constrained metabolic models for target discovery and metabolic engineering.
Human-GEM · iPC1469 · iML1515 · FBA · FSEOFAI4Science researcher
and scientific software builder
Computational biology × autonomous research

Models · Agents · Software
2026
Research profile
I work where AI4Science, research agents, and full-stack scientific software meet.
I earned a B.Sc. in Agriculture from South China Agricultural University with a GPA of 3.99/5.0 and have published four papers as a first author and collaborator.
My work turns metabolic models, protein machine learning, computer vision, and molecular simulation into workflows that are easier to inspect, repeat, deploy, and extend.
Operating fields
Genome-scale and enzyme-constrained metabolic models for target discovery and metabolic engineering.
Human-GEM · iPC1469 · iML1515 · FBA · FSEOFProtein language models, enzyme-kinetics prediction, and molecular simulation connected to biological modeling.
UniKP · UniKPNext · AlphaFold 3 · GROMACSStateful, tool-using agents that coordinate knowledge, models, code, evaluation, and research workflows.
LangGraph · MCP · RAG · Multi-Agent · HarnessReproducible training, unified inference adapters, model routing, and efficient local or cloud model serving.
PyTorch · vLLM · PEFT · Async Inference · DockerEducation & research workflows
Animal Science · Bachelor of Agriculture · GPA 3.99 / 5.0
Built, refined, and analyzed Human-GEM, iPC1469, and iML1515. Integrated enzyme-kinetics parameters into iPC1469 to construct iPC1469E and iPC1469P, then used FBA, FSEOF, COBRA, and GECKO to screen candidate genes for cordycepin production.
Fine-tuned UniKP for enzyme-kinetics prediction and connected UniKPNext with GECKO 3 modules. Built a protein–cellulose system with more than 200,000 atoms, completed a 100 ns simulation, and automated analysis with Python and MDTraj.
Built enzyme-kinetics training and evaluation pipelines with unified data splits, features, checkpoints, batch inference, and MAE / RMSE / R² comparisons. Built a cloud-and-local inference layer with routing, concurrency control, retries, streaming, and invocation logging.
Selected software systems
A LangGraph multi-agent workflow that combines LLM semantic rewriting, iterative agent evaluation, and local NLP post-processing in a state machine, with a React/Vite interface and FastAPI service.
A paper-parsing web application that converts individual or batched PDFs into RAG-ready Markdown and ships as a reproducible Docker Compose service.
An MCP server connecting MCP clients, OpenAI-compatible multimodal models, and Android devices over ADB so agents can execute mobile tasks from natural-language instructions.
Open-source contributor to xiaozhi-client ↗, RikkaHub ↗, and cscience ↗.
Peer-reviewed output
Jiang H., Guo E., Zhang Z., et al.
China Biotechnology · 46(6): 17–28Zhou Y., Zhang T., Lyu M., et al.; Jiang H.
Biotechnology Journal · 20(9): e70125Wu Z., Wang Y., Lao Y., et al.; Jiang H.
Food Bioscience · 2025: 107908Kuang J., Zhang S., Jiang H., et al.
Synthetic Biology Journal · 6(06): 1294–1310Capabilities & tools
LangGraph · Multi-Agent · State Machine · RAG · MCP · Tool Calling · Prompt Engineering · Context Engineering · Harness
GEM / GSMM · FBA · FSEOF · GECKO 3 · Enzyme Kinetics Prediction · PLM · AlphaFold 3
PyTorch · Transformers · PEFT / LoRA · vLLM · OpenAI SDK · Async Inference · Model Routing · Docker
Python · TypeScript · React · Vite · FastAPI · SSE · SQLite · Linux · Shell · Git · Pytest
Python · TypeScript / JavaScript · R · MATLAB · C / C++
COBRA Toolbox · GECKO 3 · GROMACS · Gaussian · Multiwfn · Sobtop · MDTraj · Matplotlib
React · Vite · FastAPI · Pydantic · Uvicorn · REST API · SSE · SQLite · Docker
Linux · Shell · GitHub · GitLab · uv · pnpm · Pytest · Vitest · Ruff · LaTeX · Markdown
Open to AI Agent / AI4Science / Research Engineering roles