Full-stack engineer who pairs production React/TypeScript front-ends with Python AI services in real shipping products. Two years freelancing 20+ LLM-powered apps for paying clients on Claude, OpenAI and LangChain. MSc Advanced Data Science & AI at the University of Liverpool.
I write code that other people maintain, document the workflows so non-engineers can run them, and use AI tooling daily as part of the work rather than as a side project.
const vaibhav = {
role: "Full-Stack + AI Engineer",
location: "Liverpool, UK",
status: "UK Right to Work · Graduate visa from Jan 2027",
stack: ["TypeScript", "Python", "React 19", "Next.js 15", "FastAPI"],
data: ["Postgres", "Supabase", "pgvector", "Redis", "BullMQ"],
ai: ["LangChain", "OpenAI", "Gemini", "Whisper STT", "MCP servers"],
shipping: ["MCP Marketplace", "Scholar.AI v2", "MedReviewAI"],
open_to: ["AI / Full-stack roles", "Graduate placement Jan 2027"],
};
apex · autonomous job application CLIOpens Chrome, finds LinkedIn Easy Apply roles, generates a tailored 1-page resume per job, fills custom application questions with an LLM, and submits — autonomously, until LinkedIn's daily cap. Free LLMs only (Groq · Cerebras · Gemini · Ollama). Pause-on-stuck UX, application history, ATS-safe PDFs.
|
|
|
Open-source AI agent platform. Multi-LLM router across free providers, MCP-native, local-first, cloud-deployable on Vercel + Supabase. Bring your own keys; ship agents in minutes. A Claude Cowork alternative built end-to-end on Next.js 16 + React 19.
|
|
|
The registry of Model Context Protocol servers. Browse 800+ servers, copy install snippets for any MCP-aware client, ship your own. Auto-syncs from Glama + the official MCP repo daily. Next.js 15 RSC, Cmd-K palette, dynamic OG cards, full keyboard nav.
|
|
|
AI research assistant in your browser. Summarizes any paper, chats with it, exports citations. Manifest V3, Gemini 2.5 native PDF input, side-panel mode, structured extractors for arXiv, PubMed, Nature, IEEE, bioRxiv.
|
Extract structured insights from medical research papers in seconds. PICO frame, confidence scoring, multi-source search across PubMed + bioRxiv + medRxiv. Built for clinicians and researchers who need answers, not abstracts.
|
|
AI tutor that refuses to give answers. Socratic-method LLM agent that pushes students toward the solution with hints, not handouts. Conversation memory, hint escalation, topic tracking.
|
Production literature-review pipeline. LangChain chunking, embedding, retrieval and synthesis over thousands of academic papers. Built for systematic reviews at scale, not toy demos.
|
|
Offline benchmark suite for local Ollama LLMs on Windows. Per-model latency, JSON-schema output guardrails, comparative reports. Local-first. No cloud calls, no telemetry, your hardware your numbers.
|
Personal portfolio site. Hand-coded layout with smooth scroll, glass-morphism cards, dark/light theming, responsive across breakpoints. Showcases case studies, contact, current writing.
|
Languages
Frontend
Backend & Data
AI / LLM
DevOps & Tooling
- Shipping MCP Marketplace. Public registry, auto-syncs daily · mcp-hub-registry.vercel.app
- Iterating on Scholar.AI v2. Gemini PDF, side-panel chat, library mode
- MSc Advanced Data Science & AI at University of Liverpool (Jan 2026 to Jan 2027)
- Reading: agentic eval frameworks, prompt-cache patterns, MCP server design
github.com/vaibhav4046 · 2026