- 🤖 NLP Algorithm Engineer specializing in LLMs, RAG, and Multimodal Learning.
- 🛡️ Fake News Detection Researcher focusing on BERT, Graph Attention Networks (GAT), and Contrastive Learning.
- ⚙️ Backend Developer & Maintainer of LiteLLM, focusing on high-concurrency AI gateways and model orchestration.
- 🐹 Gopher in Transition — Currently migrating core microservices from Python to Go (Gin) for enhanced performance.
- Core Models: BERT, RoBERTa, CLIP, Llama, GPT.
- Expertise: Multimodal Fusion, Semantic Retrieval, Cross-Modal Attention.
- Go (Golang) + Gin
- FastAPI
- LLM Gateway / API Proxy Design
- Docker / Microservices
- Model Routing / Load Balancing / Cost Tracking
- MS-GAT: A fake news detection framework based on Cross-Modal Similarity Perception and Graph Attention Fusion.
- CMAWSCL: Chinese Fake News Detection utilizing Weak-supervised Contrastive Learning and Cross-Modal Attention mechanisms.
- Keywords:
BERT,Graph Neural Networks,Contrastive Learning,Multimodal Alignment.
- LiteLLM (Maintainer): Actively maintaining the Python SDK and Proxy Server to call 100+ LLM APIs with unified OpenAI format, load balancing, and cost tracking.
- FULING: An end-to-end voice dialogue system allowing users to interact with AI characters via LLM-powered backends.
- aitoken-aigc-agent: A Go-based experimental ground for AIGC agents, exploring high-performance orchestration.
- 🐹 Go Ecosystem: Deepening expertise in Gin and Gorm to build robust, scalable AI backends.
- 🔍 RAG Optimization: Enhancing Dense Vector Retrieval and Reranking strategies for large-scale enterprise knowledge bases.
- ⚡ AI Gateway: Adding advanced guardrails and multi-tenant quota management to LiteLLM.
- 📧 Email: kardhbped@gmail.com
- 💬 Ask me about: NLP algorithms, LLM deployment, or the journey from Python Expert to Gopher.