I'm Sujay Gopal Mulagund, an MS student in Data Science, Analytics & Engineering at Arizona State University. I build end-to-end AI/ML systems — retrieval-augmented pipelines, LLM evaluation frameworks, and full-stack AI products, from data to deployment.
- 🎓 Currently: MS Data Science, Analytics & Engineering @ Arizona State University
- 🔭 Building: DocuMind, Model Regression Detection System, AutoHub
- 🌱 Focused on: RAG pipelines, LLM evaluation & regression testing, applied AI engineering
- 🤝 Open to: collaborations, open-source contributions, and interesting AI/ML problems
- ⚡ Status: actively building & shipping
A retrieval-augmented generation pipeline for document question-answering, running local LLMs end-to-end and containerized for portable deployment.
| Layer | Technology |
|---|---|
| Orchestration | LangChain |
| Vector Store | ChromaDB |
| LLM Runtime | Ollama |
| Containerization | Docker |
| Deployment | HuggingFace Spaces |
Detects behavioral regressions in an LLM-based customer-support email classifier when prompts or models change — a 4-axis diff detector with CI-ready exit codes, versioned prompts, and a 75-case hand-curated golden dataset.
| Layer | Technology |
|---|---|
| LLM | Gemini 2.5 Flash (via google-genai SDK) |
| Embeddings | gemini-embedding-001 |
| Eval Harness | Python |
| Testing | Pytest (16 offline tests) |
| Config | YAML (externalized thresholds) |
💻 Code (no live demo — CLI / CI tool)
A living automotive platform with a natural-language AI car assistant, side-by-side comparisons across 24 vehicles, interactive 3D models, and a live industry news feed — shipped as a pure front-end SPA.
| Layer | Technology |
|---|---|
| Frontend | React 18 + Vite + Tailwind CSS + Framer Motion |
| Routing | React Router (SPA) |
| 3D Viewer | <model-viewer> (GLB models) |
| News API | GNews.io via same-origin proxy |
| Hosting | Google Cloud Run (nginx + Docker) |
💻 Code (no confirmed live URL on hand — add one here once deployed)
Languages
Frontend
Backend & Infra
Cloud
AI & Data
