Getting Started
Welcome to the documentation for Gagan Jain’s portfolio — an AI / LLM Engineer focused on production-grade GenAI systems. This index splits the docs into two parts: Overview (how this site works) and Projects (each system’s documentation).
Overview
The Overview section covers the portfolio itself — its architecture, decision records, and a full route map.
Architecture
System design patterns used across the AI and systems projects — governance, event sourcing, hybrid retrieval, and more.
ADRs
Architecture Decision Records: why the portfolio is built on Astro 7 + Tailwind v4 + GSAP, static output on Vercel, and more.
Site Map
Every page and route on the site — where each one lives in the repo, what it does, and how it behaves.
Projects
Each project has its own documentation page — architecture, features, and the tech stack behind it.
NexusAOS
Governance-first, event-sourced AI operating environment in Rust — 12 workspace crates, 981 tests, provider-agnostic LLM streaming.
nexus-kernel
Companion to NexusAOS — the alpha-track microkernel sibling of the same 12-crate Rust workspace.
SeshaOS
Local-first AI OS with specialist models (Gemma 4, Qwen3) and LiteLLM-compatible routing.
RAG Service
Production RAG API — hybrid retrieval (dense + BM25, RRF) over ChromaDB with grounded answers and citations.
LLM Eval Harness
Golden-set evaluation: faithfulness, answer relevance, and correctness with LLM-as-judge and offline fallbacks.
Vyākṛti
Sanskrit-oriented programming language with a complete compiler pipeline and a browser-based IDE.
AIM
Production attendance platform — Flask + MySQL, Argon2id auth, Prometheus monitoring, 101 tests.
FWRS
Food-waste optimization — 3-stage lexicographic linear program (fairness → priority → cost) with expiry-aware routing.
For collaboration or opportunities, reach out via email or LinkedIn.