AI/ML Engineer working on autonomous agents, RAG, and LLM inference. MS CS @ USC · Google Cloud Certified Associate Cloud Engineer
I build AI systems that act, and I care about the layer that keeps them alive in production: inference, serving, and the Kubernetes substrate underneath.
- 🤖 AI-Image-Organizer-agent: an autonomous multimodal agent that runs local vision models through Ollama in a perception, reasoning, and action loop. Fully offline, zero cloud dependencies.
- 🧪 LLM-Finetuning: reasoning distillation with QLoRA. Fine tuning Llama 3.2 3B on reasoning traces distilled from DeepSeek R1 using Unsloth and TRL, then exporting to GGUF and serving locally with Ollama.
- 🔎 rag-agents: agentic RAG systems. A multi tool RAG agent with function calling across Wikipedia, Arxiv, and web retrievers, backed by FAISS vector search, plus advanced RAG techniques and local Llama inference.
- 🧠 Classic deep learning: emotion classification with VGG16 transfer learning at 90.33% accuracy plus RoBERTa and BiLSTM text models, facial keypoint detection, and image captioning with a CNN and RNN encoder decoder.
Benchmarking LLM serving: quantization, batching, and token throughput against GPU cost on Kubernetes. Results will be published here.
Stack: Python · PyTorch · LangChain · LangGraph · RAG · FAISS and LanceDB · Ollama · FastAPI · Kubernetes (GKE) · Docker · GCP · AWS