Full-Stack AI Developer — Build an Autonomous AI Lead Research Engine

We are building a lightweight AI-powered SaaS module that automates deep, multi-source prospective lead research. We need a skilled Full-Stack AI Engineer to build an autonomous research agent using LangGraph that can take a company domain/URL, perform web scraping, analyze data, and generate tailored outbound sales intel.

We already have a core UI prototype and a defined specification for the pipeline. Your job is to implement the LangGraph multi-agent backend, connect it to a Django / FastAPI backend service, and wire the output to our existing Next.js frontend.

This is a focused, fixed-price project designed to be straightforward and completed efficiently (~10–15 hours of focused engineering time).

What Exists Already

Frontend Shell: React / Next.js (App Router) + Tailwind CSS + Shadcn UI with mock state.

API Spec & Schema: OpenAPI specification detailing input payloads, webhook contracts, and response models.

System Prompting Guidelines: Base prompt strategies and target JSON schemas for lead intel.

What You’ll Build

LangGraph Agent Workflow

A stateful graph containing nodes for web scraping, search retrieval, information synthesis, and structured JSON output.

Logic to handle error retries, rate limits, and conditional routing (e.g., fallback logic if a site blocks scraping).

Backend API Service (Python / Django or FastAPI)

Endpoint to trigger background agent execution and stream step-by-step agent thoughts/status to the frontend (via Server-Sent Events or WebSockets).

PostgreSQL storage for research outputs and run history.

Frontend Wiring (Next.js / TypeScript)

Wire the existing Next.js frontend to display real-time execution steps, key findings, and final markdown/structured reports.

What "Done" Looks Like

A user enters a company website domain (e.g., acme.com). The LangGraph agent executes in the background, showing step-by-step agent status in the UI (e.g., "Scraping homepage...", "Analyzing press releases...", "Formatting profile..."). Within 60 seconds, the frontend populates a comprehensive lead report containing key executives, pain points, and personalized draft outreach copy.

Tech Stack & Integrations

AI Framework: LangGraph (Python or TypeScript version) + OpenAI / Anthropic API

Backend: Python (Django or FastAPI) + PostgreSQL

Frontend: Next.js (TypeScript, Tailwind CSS, React Query / TanStack Query)

External APIs: Web Search API (Tavily, SerpAPI, or Exa) + Web Scraper (Firecrawl or Playwright)

Required Skills

Strong experience building stateful multi-agent systems with LangGraph / LangChain

Solid proficiency in Python (Django/FastAPI) and TypeScript (Next.js / React)

Experience working with LLM APIs, structured outputs (Pydantic / Zod), and prompt engineering

Comfort building and consuming REST APIs / streaming protocols (SSE)

Clean code habits and good documentation standards

How We Work

Code delivered via private GitHub repository PRs

Clean separation of API keys / credentials using .env configurations

Fixed-price milestone engagement on Upwork

Milestones

Milestone 1 — LangGraph Execution Pipeline (Backend Core):

Set up the state graph with search, scraping, and synthesis nodes returning structured Pydantic data on test inputs.

Milestone 2 — API Integration & Streaming

Expose the graph via Django/FastAPI endpoints with real-time status streaming.

Milestone 3 — Frontend Wiring & Final Handover

Connect Next.js frontend, verify real-time status updates, test edge cases, and deliver clean documentation.