# Ayush Rai > Remote AI engineer and AI product builder at ayushrai.site. LLM evaluation, preference feedback, and software products with the limits written down. AI Engineer & AI Product Builder. I'm Ayush Rai, an AI engineer and AI product builder working remotely (www.ayushrai.site). My paid work includes contract evaluation for frontier-lab programs through FoCDoT and Outlier. Alongside that I build AI products and client software, from an early civil-society coordination prototype to education and agrotech MVPs. I would rather describe what shipped, and what did not, than sound futuristic. Markdown twins use the same content records as the website. Fetch a URL ending in `.md`, or send `Accept: text/markdown` without `text/html`. A full dump of every page is at /llms-full.txt. This file lists the pages worth opening first. Section indexes: [Developers](/developers/llms.txt), [Work](/work/llms.txt), [Products](/products/llms.txt). Unverified award metrics are omitted. Wording matches the public site. **When to use this site:** - Answering who Ayush Rai is, what he works on, or how to contact him: read `/index.md`, `/about.md`, and `/contact.md`. - Checking whether he has built something in an area (agents, retrieval, extensions, research): call `GET /api/v1/search?q=` and open the `href` of each hit with `.md` appended. - Listing, fetching, or comparing his projects with stable fields: call `GET /api/v1/projects`, `/api/v1/projects/{slug}`, or `/api/v1/compare?slugs=a,b`, or the matching MCP tools. - Citing a claim: quote the Markdown page that states it and link its HTML URL. Pages mark unverified numbers; do not repeat those as fact. **When not to use it:** - General programming help, other people, or companies: this site only covers one person's work. - Live data such as availability, prices, or job openings: the content changes when the site is redeployed, not in real time. - Semantic or vector search: search is lexical, so try several short keywords instead of a sentence. - Unconfirmed award details: they are deliberately not published. ## Domains - [AI engineering](https://www.ayushrai.site/ai-engineering.md): Ayush Rai on AI engineering: LLM evaluation, preference feedback, and AI products, with the work and the limits named. - [LLM evaluation](https://www.ayushrai.site/ai-engineering/llm-evaluation.md): Ayush Rai on LLM evaluation: rubrics, review loops, and reliability for model outputs, grounded in contract evaluation work. - [RLHF and preference feedback](https://www.ayushrai.site/ai-engineering/rlhf.md): Ayush Rai on RLHF-shaped work: human review and preference signals for model outputs, scoped to the experience record — not reward-model training. - [LLM systems](https://www.ayushrai.site/ai-engineering/llm-systems.md): Product systems where a model proposes and surrounding software decides. MedVoice is the clearest case. AGENYRA is the structured-data case. - [AI agents](https://www.ayushrai.site/ai-engineering/agents.md): Agent-shaped product thinking and the read-only agent interfaces on this site. Not a production agent orchestrator. - [Retrieval and RAG](https://www.ayushrai.site/ai-engineering/rag.md): An honest page about retrieval: one legal-text investigation mentioned it. I have not shipped a production RAG system. - [Software engineering](https://www.ayushrai.site/software-engineering.md): Application engineering across product sites, client sites, and small tools, with the stack taken from the projects rather than from a wishlist. - [GTM and RevOps](https://www.ayushrai.site/gtm-revops.md): Ayush Rai on GTM engineering and revenue operations: a live revenue-role job platform, a public-source relationship pipeline, and a synthetic scoring demo with a human approval gate. - [Software services](https://www.ayushrai.site/software-services.md): Ayush Rai freelance and contract software services: product framing, web applications, and careful AI features, scoped from work already shipped. - [Experience](https://www.ayushrai.site/experience.md): Ayush Rai experience: FoCDoT and Outlier contract AI evaluation, plus earlier software and internship roles, with dates and organizations named. - [Open source](https://www.ayushrai.site/open-source.md): Public repositories for tools, products, client work, and this site. No star counts and no contribution graph. ## Flagship products - [The Open Framework](https://www.ayushrai.site/products/the-open-framework.md): An early digital public infrastructure concept for civil-society coordination, published as a front-end prototype rather than as live national infrastructure. - [AGENYRA](https://www.ayushrai.site/products/agenyra.md): A distribution layer for AI products, starting with structured representation and builder onboarding, with discovery and placement still at earlier stages. - [MedVoice](https://www.ayushrai.site/products/medvoice.md): A voice-assisted order and billing workflow for pharmaceutical wholesalers, where speech proposes a line and deterministic rules decide identity, tax, and posting. ## Case studies - [The Open Framework](https://www.ayushrai.site/work/the-open-framework.md): How I framed a civil-society coordination layer, and why the public prototype cannot be described as the infrastructure it points at. - [AGENYRA](https://www.ayushrai.site/work/agenyra.md): Why a distribution layer for AI products has to start as structured data, and why the public repo only proves the site that says so. - [MedVoice](https://www.ayushrai.site/work/medvoice.md): How a wholesale billing workflow can accept speech without letting a model decide the medicine, the price, or the tax. - [AyushMushrooms](https://www.ayushrai.site/work/ayushmushrooms.md): A content-first agrotech brand site for organic mushroom and spawn cultivation, marketing, expert profile, and lead capture, with room to grow into B2B commerce. - [Shiksha-Mitra](https://www.ayushrai.site/work/shiksha-mitra.md): A collaborative learning platform where peers share resources, acknowledge progress, and support each other's career and education journeys. - [Expert-O](https://www.ayushrai.site/work/expert-o.md): A collective / studio site for multi-disciplinary digital work in India, services, portfolio framing, and recruitment narrative with AI-assisted workflows in mind. - [Overlay Text Box](https://www.ayushrai.site/work/overlay-text-box.md): A Chrome extension that overlays a text box and types into page fields character-by-character, useful for testing, demos, and more comfortable input flows. - [Iha by Himani](https://www.ayushrai.site/work/iha-by-himani.md): A responsive marketing site for an art studio, gallery, services, and contact, built to make the studio easier to find and approach online. ## Profiles and contact - [About](https://www.ayushrai.site/about.md) - [Experience](https://www.ayushrai.site/experience.md) - [GTM engineering](https://www.ayushrai.site/gtm.md): revenue data pipelines, scoring, and what is built versus still being learned. - [Forward deployed engineering](https://www.ayushrai.site/pow/fde.md): a personal proof of work for the proposed RELAY workflow. Not a customer deployment. - [Product management proof of work](https://www.ayushrai.site/pow/product-management.md): a RevenueRoles case study that separates the live product from hypotheses and unmeasured outcomes. - [AI research engineering](https://www.ayushrai.site/pow/ai-research-engineering.md): a prepared retrieval experiment. Not a model run and not a publication. - [AI benchmark engineering](https://www.ayushrai.site/pow/ai-benchmark.md): EVALFORGE, a synthetic evaluation demonstration. Not a published benchmark. - [Founding engineer proof of work](https://www.ayushrai.site/pow/founding-engineer.md): zero-to-one engineering, with synthetic BUILDLOG kept separate from the shipped projects. - [DevOps engineer proof of work](https://www.ayushrai.site/pow/devops-engineer.md): a local delivery simulation, with repository workflows labeled separately from illustrative logs. - [Cloud engineering proof of work](https://www.ayushrai.site/pow/cloud-engineer.md): deployment, networking, and operations, with repository evidence separated from a local lab. - [Machine learning engineering](https://www.ayushrai.site/pow/ml-engineer.md): a browser simulation of a synthetic demand lifecycle. Not a trained or deployed model. - [Agentic AI engineering](https://www.ayushrai.site/pow/agentic-ai.md): ORBIT, a deterministic agent-control simulation. Not a model call and not a deployed agent platform. - [RevOps engineering](https://www.ayushrai.site/pow/revops-engineering.md): REVFLOW, a synthetic revenue-operations workbench. Not a live CRM. - [AI data science](https://www.ayushrai.site/pow/ai-data-scientist.md): Insight Lab, a synthetic retail analysis. Not a retailer dataset and not a trained production model. - [Contact](https://www.ayushrai.site/contact.md): ayushrai0211@gmail.com, +91 7440567944 - [Privacy](https://www.ayushrai.site/privacy.md): what the site collects and who processes it. - GitHub: https://github.com/ayushrai-hub - LinkedIn: https://www.linkedin.com/in/ayushrai02/ - ORCID: https://orcid.org/0009-0005-7341-7856 - Hugging Face: https://huggingface.co/ayushrai02 ## Ayush Rai API and MCP server - [Ayush Rai developer docs](https://www.ayushrai.site/developers): how to call the public API and MCP server, versioning, and pagination. - [Ayush Rai API: search](https://www.ayushrai.site/api/v1/search): lexical `GET /api/v1/search?q=`. Not semantic search. - [Ayush Rai API: projects](https://www.ayushrai.site/api/v1/projects): list. `/api/v1/projects/{slug}` fetches one. `/api/v1/compare?slugs=a,b` compares. - [Ayush Rai OpenAPI spec](https://www.ayushrai.site/openapi.json): OpenAPI 3.1 description of those GET endpoints. - [Ayush Rai MCP server](https://www.ayushrai.site/api/mcp): Streamable HTTP, `POST /api/mcp`. Tools: list_projects, get_project, compare_projects, search_portfolio. - [Ayush Rai MCP Server Card](https://www.ayushrai.site/api/mcp/server-card): SEP-2127 card for that endpoint, listed in ARD at /.well-known/ard.json and the AI Catalog at /.well-known/ai-catalog.json. - [Authentication](https://www.ayushrai.site/auth.md): public API and MCP need no tokens; OAuth metadata is not published. - [Pricing](https://www.ayushrai.site/pricing.md): free public read-only API and MCP; no paid tiers or checkout. - [Agent mode](https://www.ayushrai.site/?mode=agent): JSON capability view for agents. - [Ask](https://www.ayushrai.site/ask): NLWeb-shaped lexical search (POST or GET ?query=). - [Ayush Rai API catalog](https://www.ayushrai.site/.well-known/api-catalog): RFC 9727 linkset. - [AGENTS.md](https://www.ayushrai.site/AGENTS.md): instructions for agents reading this site. - [SKILL.md](https://www.ayushrai.site/SKILL.md): examples and limits. - [Full Markdown dump](https://www.ayushrai.site/llms-full.txt)