# Mayank Saini

> Mayank Saini is an AI engineer in India. His work spans LangGraph agent orchestration, MCP tools, context management and financial-close data platforms. His machine-learning research studies multimodal tool orchestration and learned model routing. This is his personal engineering portfolio.

Mayank works at Mirian AI and previously worked in PwC US R&D. Relay is his personal product prototype, not a released production platform. Public research links are arXiv preprints; study results must retain their evaluation context. The portfolio guide answers from the curated public documents below.

This file is an optional public context export for readers and tools.

## Public profile
- [Engineering profile](https://mayanksaini.dev/index.md): Professional work, education, research, interests and contact details.
- [Relay prototype](https://mayanksaini.dev/relay.md): Built mockup, implementation underway, architectural decisions and interactive portfolio demonstration.
- [Field Notes sample](https://mayanksaini.dev/reading/): A dated public sample of a personalized local reading desk, customized from ThariqS/ai-newtab.
- [Current one-page CV](https://mayanksaini.dev/Mayank-Saini-CV.pdf): Downloadable career summary.

## Research and code
- [One Supervisor, Many Modalities](https://arxiv.org/abs/2603.11545): 2026 preprint on adaptive tool orchestration.
- [Towards Resource-Efficient Multimodal Intelligence](https://arxiv.org/abs/2511.06441): 2025 preprint on learned model routing.
- [Agent Expression](https://github.com/itsMay21/agent-expression): Portable skills for agent-written text and code.
- [GitHub](https://github.com/itsMay21): Public code profile.