Ziqian Bi 毕梓仟

Ph.D. in Computer Science · ECE/EE Background 计算机科学博士 · 电子与电气工程背景
Purdue University (bi32@purdue.edu) · Indiana University (bizi@iu.edu)
Google Scholar [PDF] Resume GitHub LinkedIn

I am a Ph.D. candidate working across Large Language Models (LLMs), multimodal Artificial Intelligence (AI), medical AI, sensing security, embedded hardware, sensing platforms, intelligent manufacturing systems, Machine Learning (ML) systems, and end-to-end AI deployment. Target roles: Research Engineer, Applied Scientist, ML Engineer, LLM / AI Systems, hardware AI, and intelligent manufacturing systems roles. My background also combines model development, system optimization, retrieval pipelines, embedded integration, hardware-aware experimentation, and bilingual communication for AI products, intelligent systems, manufacturing, and research.

Site Guide

Experience

Full CV-style page with research interests, technical skills, industry experience, research experience, and education.

Projects

Ranked GitHub repository index, from the most job-relevant AI / systems work to the full repository archive.

Publications

Representative publications first, followed by the complete publication list formatted for reading.

Current Focus

  • LLM evaluation, reasoning, and post-training
  • Multimodal and medical AI systems
  • Security and robustness for sensing and model pipelines
  • From-scratch learning projects and systems implementation

Research Interests

Large Language Models (LLMs)Multimodal Large Language Models (MLLMs)Natural Language ProcessingGenerative ModelsDiffusion ModelsComputer VisionProcess Defect DetectionYield InspectionProduction-Line OptimizationMedical AIModel Security and RobustnessModel InterpretabilitymmWave SensingSensorsEmbedded Systems

Technical Skills

Programming Languages
PythonCC++CUDAMATLABSQL
Frameworks
PyTorchTensorFlowDeepSpeedTransformersDiffusersLangChainLangGraphPydantic-AIFastAPIDjango
Specialized Models
Large Language Model (LLM) / Multimodal Large Language Model (MLLM)Diffusion Transformer (DiT) / Stable DiffusionCLIP / ViT / ResNetYOLORetinaFacePoint TransformerSimCLR3D VQGAN
Tools and Platforms
DockerVertex AIOllama / vLLMFAISSRAG systemsPostgreSQLGit
Hardware
mmWave radar (TI IWR6843 / 1843, Cascade EVM)STM32ESP8266 / ESP32I2C / SPI / UART / RS-485 / Modbusoscilloscopelogic analyzerdigital LCR meterwaveform generatorhot-air rework stationGPU / CUDA optimization
Libraries
NumPyPandasScikit-learnXGBoostAutoGluonSHAPOpenCVVTKMayavi

Selected Highlights

  • Improved BERT validation measurements from 10^-6 to 10^-12 at Xperf through automated optimization.
  • Built full-stack LLM and retrieval systems supporting 500+ queries per second during my Infraeo internship.
  • Published or contributed to work spanning ICDCS, ICCCN, Journal of Biomedical Informatics, BIO Integration, Computation, and multiple domain journals.

AI-Augmented Engineering Scope

This is not Vibe Coding or casual exploration. It describes what I can build with Agentic AI tools through Test-Driven Development (TDD), Spec-Driven Development (SDD), Verification, reproducible implementation, Debugging, and maintainable production-grade engineering. The Technical Skills section is a conservative baseline for technologies I can understand, inspect, debug, modify, and maintain directly; with Agentic AI assistance, the practical implementation scope becomes broader, while my role shifts toward Specification, Verification, Integration, and Engineering Review rather than memorizing every implementation detail.

AI Coding and Agent Tools
Claude CodeQwen CodeCodexGitHub CopilotGemini CLIAntigravityCursorGitHub Spec Kit
Agent and Large Language Model (LLM) Application Architecture
LangChainLangGraphPydantic-AIPydantic Graph
API and Real-Time Communication
GraphQLgRPCtRPCRESTful APISwagger / OpenAPIWebRTCWebSocket
Large Language Model (LLM) Deployment and Model Runtime
OpenRouterOpenAI APIOllamavLLMHugging Face TransformersHugging Face Diffusers
Large Language Model (LLM) Fine-Tuning and Alignment
LoRAQLoRAUnslothDPOGRPOSFT
Neural Network and Vision Architectures
ResNetVGGYOLOViTCLIPBLIPSAMPoint TransformerDiffusion Transformer (DiT)Latent Diffusion Model (LDM)U-NetTransformerMambaRWKVMoEMasked Autoencoder (MAE)Vision MambaSwin TransformerConvNeXtNeRFSLAM3D Gaussian Splatting
Foundation Models and LLM Families
LlamaDeepSeekQwenKimiGPTClaudeGeminiMistralMixtralGLMInternLMYiPhiGemmaStable DiffusionMiniMaxNemotronLLaVAQwen-VLInternVLDeepSeek-OCRGLM-OCR
Acceleration and AI Compute
CUDACANNMPSMulti-processingDeepSpeedRay
Medical Imaging and Neuroimaging Tools
FreeSurferFSLSimpleITKNiBabelMONAI
AI-Augmented Data Science and MLOps
PolarsSciPyOptunaMLflowStreamlitGradioGrafanaPrometheus
Machine Learning Methods
Linear RegressionLogistic RegressionSupport Vector Machine (SVM)k-Nearest Neighbors (k-NN)Naive BayesDecision TreeRandom ForestGradient BoostingXGBoostLightGBMCatBoostPCAt-SNEUMAPK-MeansDBSCAN
AI-Assisted Learning and Development
Multi-Agent AI LearningRapid PrototypingImplementation ComparisonSystem DebuggingProduction-Style Full-Stack Projects
Frontend
HTMLJavaScriptCSSTailwind CSSBootstrapReactVueReact Native
Backend
FastAPIFlaskDjango
Messaging and Streaming
KafkaRabbitMQ
Databases
MySQLMongoDBNeo4jRedisPostgreSQL
Containers and Orchestration
KubernetesDockerNginxngrok
Embedded and IoT
Raspberry PiOrange PiSTM32ArduinoESP8266 / ESP32SensorsRS-485ModbusPLCFPGACPLD
Cloud and Engineering Workflow
GCPAWSTerraformGitGitHub ActionsTest-Driven DevelopmentSpec-Driven DevelopmentTrunk-Based DevelopmentGitHub FlowGit Flow
About This Website Maintenance Docs Source

Tech Stack: Astro, Tailwind CSS, PhotoSwipe, pnpm, and MkDocs.

CI/CD and Quality Gates: Prettier, Stylelint, Playwright, GitHub Actions, and Spec-Driven Development (SDD).

AI-Friendly Engineering: Complete Specs, Clear Structure, Layered Agent Guidance, and Context-Efficient Maintenance.

Local Assets and No CDN Dependency for Stronger Network Availability.