Personal portfolio for Shiv Kumar, a Lead Engineer building production-grade GenAI, RAG, Agentic AI, realtime voice AI, and applied ML systems.
Live site: ms-kumar.github.io/shivkumar-portfolio
Profiles: LinkedIn · GitHub · Resume
Shiv works across:
- Production GenAI agents for finance automation and FP&A analysis
- Realtime voice AI with OpenAI Realtime API, Twilio, and live human handoff
- Regulatory RAG with hybrid retrieval, reranking, grounded citations, and offline deployment
- Computer vision, autonomous driving, BEV perception, and accelerator optimization
The portfolio includes detailed work from HCL Technologies, PathPartner Technology (KPIT), and AgVerse Technology, along with publications in autonomous driving and computer vision.
- 5 years of hands-on AI and ML engineering experience
- 80% retrieval hit rate and 75.8% nDCG on regulatory search
- 75% CPU runtime reduction on Qualcomm edge deployment
- Published research in Discover Artificial Intelligence and IEEE ICRAIS
- Best Paper recognition at IEEE ICRAIS 2023
Python, C/C++, FastAPI, Google ADK, Vertex AI, Agent Engine, OpenAI, Milvus, Redis, MongoDB, Docker, GCP, PyTorch, TensorFlow, OpenCV, vLLM, BM25, ModernBERT, ONNX, and Linux.
Requirements: Hugo Extended 0.162.0, Go 1.21+, Node 22, and pnpm.
pnpm install
hugo server -DOpen http://localhost:1313/shivkumar-portfolio/.
Production build:
hugo --gc --minifyGitHub Actions builds and deploys the site to GitHub Pages from main.
The site code is MIT licensed; see LICENSE.md. Resume content, personal information, photographs, and project descriptions are personal content and should not be reused without permission.