portfolio

The work.

Public, MIT-licensed artifacts on the work I want to be doing more of. Generative UI, agent orchestration, eval rigor, MCP server design. Live demos, hosted with rate limits and budget caps. Read the source.

Open source on user-facing AI surfaces8 entries

tablesalt — CSV agent with generative UI, reasoning trace, and live eval scoreboard
Public open-source data-exploration agent. Drop a CSV, ask a natural-language question, see generative UI — five render kinds (table, bar, line, stat, list) chosen by the model. text-to-SQL via the Vercel AI SDK + Vercel AI Gateway over DuckDB-WASM running in-browser; zero backend. The agent emits a 4-step reasoning trace before its answer; the eval scoreboard (12 labeled NYC-311 cases, render-kind / SQL-executes / SQL-semantic-match scored with live cost + latency) runs on demand from the front page. Consumes streamfield@^0.1.0 from npm for its streaming reasoning UI.
Author · Product EngineerNext.js 16React 19TypeScriptTailwind v4Vercel AI SDKVercel AI GatewayDuckDB-WASMstreamfield
streamfield — React primitive for partial-object stream UIs
Public open-source React primitive (~150 LOC) for rendering Vercel AI SDK partial-object streams with field-by-field reveal physics. Diffs successive snapshots, derives per-field pending → streaming → complete state, hands the state to children via render prop. Three opinionated CSS variants ship in the package; consumers can ignore them and style state transitions themselves via a data attribute. Published to npm; extracted from tablesalt. Docs site at streamfield.kevinmurphywebdev.com runs on Next.js 16 deployed to Vercel.
Author · Product EngineerTypeScriptReacttsupVercel AI SDKVercelNext.js 16
fedbench — LLM eval harness for grounded Q&A
Public open-source evaluation harness for grounded Q&A agents over policy PDFs. Two side-by-side federal corpora (Medicare + OSHA, 21 verified Q&A pairs total), BM25 retrieval, fallback ladder, deterministic citation-check + Opus 4.7 LLM-as-judge, and a no-API-key replay path so visitors can demo the scoring pipeline in under a second.
Author · Applied AITypeScriptBunAnthropic SDKMCPBM25LLM-as-judge
fieldops-mcp — agent tool server for a field-services workflow
Public open-source MCP server that exposes a small-business field-services dispatcher workflow — triage, scheduling, customer comms, utilization roll-up, human escalation — as six agent tools an LLM client can drive end-to-end. Each tool exercises a distinct shape a product engineer integrating AI has to design for: read, search, mutation with typed conflict errors, composition, aggregation, and refusal/escalation.
Author · Applied AITypeScriptBunMCP TS SDKZodClaude Desktop
grant-pilot — federal-grants agent with sub-agent orchestration
Public open-source agent that helps a small business or nonprofit discover federal grants they qualify for and drafts an application skeleton — orchestrating three specialist sub-agents (discovery, eligibility, drafter) over live federal opportunity and entity-registration APIs, with per-sub-agent fallback ladder, structured-failure routing, and a hosted demo hardened with a 5-intent allowlist, daily budget cap, and per-IP rate limit.
Author · Applied AITypeScriptBunAnthropic SDKNext.js 16VercelUpstash
kev-o — grounded RAG chatbot trained on my own writing
Public open-source grounded chatbot answering questions about my work using only my public corpus (blog, case studies, resume, OSS READMEs). Hybrid retrieval — BM25 candidate pool plus Voyage cross-encoder rerank — feeds Claude Sonnet 4.6 streamed through the Vercel AI SDK. Three surfaces share one brain: a standalone subdomain, a global Command-K palette, and inline punch-ins at the foot of every curated entry. Hardened with per-IP rate limit, daily USD spend cap, and an owner-bypass route with timing-safe key comparison and per-IP attempt cap.
Author · Applied AITypeScriptNext.js 16Vercel AI SDKAnthropic SDKVoyage rerankBM25UpstashTailwind v4
mdx-corpus — npm package: MDX directory to retrieval-ready corpus
Public npm package that turns a directory of MDX files into a retrieval-ready JSON corpus for RAG pipelines. Parses frontmatter, strips JSX components while preserving their text children, chunks on heading boundaries with a paragraph-split fallback for long sections, emits passages carrying source URL and metadata for citation. ~300 lines of TypeScript, 19 tests, dual ESM/CJS build via tsup, zero runtime dependencies. Extracted from the kev-o-ai-search build because the parse-and-chunk step is reusable across surfaces; the package deliberately refuses to grow into embeddings, vector storage, or retrieval logic.
Author · Applied AITypeScripttsupvitestMDXnpm
kevinmurphywebdev.com — personal site + portfolio + resume
Personal site, portfolio, blog, and live resume — Next.js 16 App Router with Server Components, MDX-driven case studies and blog posts, programmatic resume rendered from a single JSON source of truth, and a per-build PDF export pipeline (Puppeteer over the rendered /resume route) so the downloadable PDF can never drift from the live site. Lighthouse 95+ across all four scores; deployed on Vercel.
Author · designer · engineerNext.js 16React 19TypeScriptTailwind v4MDXVercel

Looking for the production-at-federal-scale work? 6 entries here →

tablesalt — CSV agent with generative UI, reasoning trace, and live eval scoreboard

streamfield — React primitive for partial-object stream UIs

fedbench — LLM eval harness for grounded Q&A

fieldops-mcp — agent tool server for a field-services workflow

grant-pilot — federal-grants agent with sub-agent orchestration

kev-o — grounded RAG chatbot trained on my own writing

mdx-corpus — npm package: MDX directory to retrieval-ready corpus

kevinmurphywebdev.com — personal site + portfolio + resume