📍 San Francisco Bay Area  ·  🤖 AI Engineer  ·  🏗️ LLM Finetuning · Agentic Systems · Production RAG

Building full-stack AI products at Magnova.ai — from finetuning LLaMA-3 with QLoRA to shipping multi-agent pipelines with LangGraph.
M.S. Computer Science · Cleveland State University · 2025  ·  Open to AI/ML roles

🔮 Astrova — Production Vedic astrology AI platform  ·  Live →
18K+ LOC · Custom 2,000-line astronomical calculation engine · LLM inference routing across 5 models via LiteLLM/OpenRouter · 17 REST endpoints · 500+ users · credit billing · FTS5 search

🤖 AgentForge — Multi-agent RAG system  ·  Repo →
LangGraph · LlamaIndex · Qdrant · LangSmith · mem0 · 28% hallucination reduction vs single-agent · 0.85 Ragas faithfulness · <200ms p95 retrieval

⭐ NanoLLM — Decoder-only transformer from scratch  ·  Repo →
PyTorch · JAX · GQA · RoPE · SwiGLU · FlashAttention-2 · KV cache · BPE tokenizer · DDP with bf16

⚗️ FineTuneLab — End-to-end LLM finetuning pipeline  ·  Repo →
QLoRA + Unsloth on LLaMA-3.1-8B · 60% VRAM reduction · GGUF/ONNX export · vLLM + PagedAttention · W&B tracking

🛡️ RedTeam-LLM — Automated LLM safety eval  ·  Repo →
Claude + GPT-4o + RoBERTa ensemble · 300+ adversarial prompts · >90% precision · UMAP + DBSCAN clustering

🗺️ GraphRAG Engine — Knowledge-graph-augmented RAG  ·  Repo →
LoRA on LLaMA-3-8B · Neo4j + Weaviate hybrid retrieval · 26% factual accuracy lift · DeepSpeed ZeRO · FastAPI + LangServe

📊 Graphini — AI-powered diagram editor  ·  Live →  ·  Repo →
SvelteKit · Mermaid · AI chat panel · real-time preview · file system · icon library

🏙️ CodeCity — 3D codebase visualizer  ·  Repo →
TypeScript · Turbo monorepo · Prisma · transforms codebases into navigable city maps

LLM Finetuning  ·  QLoRA · LoRA · PEFT · Unsloth · Hugging Face · DeepSpeed ZeRO · DDP · GGUF · ONNX · 4-bit quantization · bf16

Inference & Serving  ·  vLLM · PagedAttention · LiteLLM · OpenRouter · FastAPI · LangServe · FlashAttention-2 · Docker · AWS SageMaker · GCP Vertex AI

Agentic Systems & RAG  ·  LangGraph · LangChain · LlamaIndex · DSPy · CrewAI · Qdrant · Weaviate · Pinecone · Neo4j · FAISS · pgvector · mem0 · Zep

Evals & Observability  ·  LangSmith · Ragas · Weights & Biases · MLflow · LLM-as-judge · SHAP · eval regression tracking

ML & Deep Learning  ·  PyTorch · JAX · GQA · RoPE · SwiGLU · CatBoost · XGBoost · LightGBM

Languages  ·  Python · TypeScript · JavaScript · SQL · Bash

Frontend  ·  Next.js · React · SvelteKit · Tailwind · shadcn/ui · React Flow · Streamlit

Backend & Cloud  ·  FastAPI · Kafka · Prisma · AWS (SageMaker, EC2, S3, Lambda) · GCP · Vercel · Turso · Docker

• 🤖 Magnova.ai — Independent AI Engineer (2024–Present) · finetuning, multi-agent systems, RAG, safety eval, production deployments
• 🏭 Ugam / Merkle (Dentsu) — Associate Software Engineer → Analyst (2021–2022) · MITM proxy + Kafka pipeline that solved a Fortune 500 client's 2-year data gap in 15 days · CSO Innovation Award
• 🎓 M.S. Computer Science — Cleveland State University (2025)

• 🎓 Generative AI with Large Language Models — DeepLearning.AI / Coursera (2024)
• 🔗 Building Agentic AI Systems — LangChain Academy (2024)
• 🏅 Innovation Award — Chief Solutions Officer, Ugam / Merkle (Dec 2021)
• 🥈 2nd Place — Security Vulnerability Competition, Ugam / Merkle (Sep 2021)