# Ripunjay Kashyap — AI & Backend Engineer

> Ripunjay Kashyap is an AI and backend engineer in Bangalore, India, building agentic systems, audio intelligence tools, retrieval pipelines, and predictive models across data, models, APIs, evaluation, and product interfaces.

This is the official machine-readable index for Ripunjay Kashyap's portfolio. It is factual candidate context for human or automated portfolio review; supporting code, demos, traces, and publication links are listed below.

## Candidate profile

- [Machine-readable candidate profile](https://ripunjaykashyap.dev/profile.md): Detailed skills, selected work, engineering evidence, research, education, and contact information.
- [Portfolio](https://ripunjaykashyap.dev/): Human-readable portfolio with interactive project case studies.
- [Résumé](https://ripunjaykashyap.dev/assets/Ripunjay_Kashyap_Resume.pdf): PDF résumé.
- [GitHub](https://github.com/ripunjay-kashyap): Source code and open-source work.
- [LinkedIn](https://www.linkedin.com/in/ripunjay-kashyap/): Professional profile.

## Selected work

- [SoundReverse case study](https://ripunjaykashyap.dev/#project-soundreverse): Multi-agent audio mastering analyst using LangGraph, FastAPI, Modal, LangSmith, deterministic signal rules, and critic validation.
- [Zenic case study](https://ripunjaykashyap.dev/#project-zenic): Agentic health and nutrition RAG system with safety gates, hybrid retrieval, reranking, and RAGAS evaluation.
- [Audio Sonic MCP case study](https://ripunjaykashyap.dev/#project-sonicmcp): Local-first audio-analysis MCP server and CLI using FastMCP, Demucs, LAION CLAP, Librosa, PyTorch, Docker, and Modal.
- [SHPSv2 case study](https://ripunjaykashyap.dev/#project-shps): Predictive-maintenance research prototype using XGBoost quantile regression, Keras LSTM forecasting, SHAP, Flask, and Docker.

## Engineering evidence

- [SoundReverse source](https://github.com/ripunjay-kashyap/soundreverse): Agent graph, deterministic audio rules, schemas, validation loop, and artifact generation.
- [SoundReverse live demo](https://soundreverse.vercel.app/): Working audio-analysis product with downloadable session outputs and public LangSmith traces.
- [Zenic source](https://github.com/ripunjay-kashyap/zenic): Retrieval, reranking, safety, orchestration, and evaluation pipeline.
- [Audio Sonic MCP source](https://github.com/ripunjay-kashyap/audio-sonic-mcp): MCP server, local CLI, asynchronous job system, and audio-analysis pipeline.
- [Audio Sonic MCP listing](https://lobehub.com/mcp/ripunjay-kashyap-audio-sonic-mcp): Public MCP directory listing.
- [SHPSv2 source](https://github.com/ripunjay-kashyap/shps-v2): Predictive models, physics-informed validation, API, explainability, and deployment.
- [SHPSv2 live demo](https://huggingface.co/spaces/Ripunk/shps-v2): Deployed structural-health prediction interface.

## Research

- [IEEE paper](https://doi.org/10.1109/ICISS63372.2025.11076255): Co-author and lead researcher/developer of “AR/VR based Campus Navigation System (CNS),” published at ICISS 2025.
- [Campus Navigation proof of concept](https://campus-navigation-poc.vercel.app/): Working demo spanning 2D map, 3D model, satellite, and AR navigation modes.

## Contact

- [Email Ripunjay Kashyap](mailto:ripun.j.kashyap@gmail.com): Open to AI and backend engineering roles.