Computer Science student exploring the intersection of AI systems, machine learning, and backend engineering.
I enjoy building things from first principles—from implementing tokenizers and neural network components to designing agentic workflows, distributed systems, and multimodal deep learning architectures. My work spans AI evaluation pipelines, LLM-powered automation, backend infrastructure, and applied machine learning research.
Currently building agentic ML systems, developing scalable backend services, and researching multimodal deepfake detection using cross-attention and temporal transformers.
| Repository | Stars | Pull Request | Status |
|---|---|---|---|
| Hugging Face ML Intern | ⭐ 10k+ | Handle ImageContent in MCP tool results, pass images to vision-capable LLMs (#262) | Open 🔄 |
| AgentScope | ⭐ 27k+ | refresh read file cache recency on access (#1811) | Open 🔄 |
| AgentScope | ⭐ 27k+ | Support windows-style separators in glob patterns (#1809) | Open 🔄 |
Contributing to high-impact open-source AI frameworks and agent systems.
-
HyperAttention: Scaling Transformer Attention Beyond Quadratic Complexity
-
Adversarial Corpus Injection: How 250 Poisoned Samples Compromise LLMs Across All Parameter Scales
-
ReLU and the Geometry of Non-Linearity: Why Stacked Linear Layers Aren't Enough
-
Beyond REST: Understanding gRPC's Role in High-Performance Microservices Architecture
-
Supply Chain Attacks Explained: Inside the LiteLLM Credential-Harvesting Backdoor