Backer is a Next.js 14 platform that connects founders and investors through short-form pitch content, profile matching, messaging, and investment intent workflows.

• Next.js 14 (App Router) + React 18 + TypeScript
• NextAuth (credentials provider)
• DynamoDB (primary app data store)
• AWS S3 (media uploads)
• AWS SES (optional invite email delivery)
• Prisma (present in repo for schema/client tooling and legacy scripts)

• Founder and investor registration/login
• Founder onboarding/profile + startup/company creation
• Investor onboarding/profile + reels-style feed (/investor/feed)
• Product like/interest and commitment flow
• Founder/investor messaging
• Co-founder invite flow

• Node.js 18+ (recommended: Node.js 20)
• npm
• AWS account/credentials with access to:
  • DynamoDB
  • S3
  • SES (only if you want invite emails)

Create a .env file in the project root. At minimum, configure:

# NextAuth
NEXTAUTH_SECRET="replace-with-a-strong-random-secret"
NEXTAUTH_URL="http://localhost:3000"

# AWS
AWS_REGION="us-east-2"
AWS_PROFILE="your-profile" # or AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY

# S3 uploads
AWS_S3_BUCKET="your-s3-bucket"
AWS_S3_PUBLIC_URL="https://your-bucket.s3.us-east-2.amazonaws.com"
AWS_S3_UPLOAD_PREFIX="uploads"

# Optional S3 custom endpoint/localstack
# AWS_S3_ENDPOINT="http://localhost:4566"
# AWS_S3_FORCE_PATH_STYLE="true"

# Optional SES (co-founder invite emails)
EMAIL_FROM_ADDRESS="no-reply@your-domain.com"
APP_BASE_URL="http://localhost:3000"

# DynamoDB table names
DYNAMODB_TABLE_USERS="backer-dev-users"
DYNAMODB_TABLE_FOUNDERS="backer-dev-founders"
DYNAMODB_TABLE_PRODUCTS="backer-dev-products"
DYNAMODB_TABLE_FOUNDER_PRODUCTS="backer-dev-founder-products"
DYNAMODB_TABLE_FOUNDER_INVITES="backer-dev-founder-invites"
DYNAMODB_TABLE_INVESTORS="backer-dev-investors"
DYNAMODB_TABLE_INVESTOR_INTERESTS="backer-dev-investor-interests"
DYNAMODB_TABLE_FEED_EVENTS="backer-dev-feed-events"
DYNAMODB_TABLE_CONVERSATIONS="backer-dev-conversations"
DYNAMODB_TABLE_MESSAGES="backer-dev-messages"
DYNAMODB_TABLE_INVESTMENTS="backer-dev-investments"

# Optional ML model output path
# FEED_MODEL_PATH="lib/feed/model/feed-model.json"

# Optional local DynamoDB endpoint
# AWS_DYNAMODB_ENDPOINT="http://localhost:8000"

Notes:

• Use either AWS_PROFILE or static access keys, not both.
• Do not commit real secrets in .env.

1. Install dependencies:

npm install

2. Ensure DynamoDB tables exist:

npm run setup:dynamodb

3. Start the app:

npm run dev

4. Train the feed model (recommended after generating some usage data):

npm run train:feed-model

5. Open:

http://localhost:3000

• npm run dev - start local dev server
• npm run build - production build
• npm run start - start production server
• npm run lint - run ESLint
• npm run setup:dynamodb - create required DynamoDB tables if missing
• npm run train:feed-model - train and persist feed ranking model weights

The investor reels feed uses a trainable logistic ranking model. At request time, each candidate startup is featurized and scored, then sorted by predicted relevance.

• Stage match between investor preference and startup stage
• Investor stage affinity learned from prior investor actions
• Text affinity between investor history/tags and startup content
• Interest tag overlap
• Global product popularity from likes/commitments
• Freshness decay

• Training script: scripts/train-feed-model.ts
• Model artifact output: lib/feed/model/feed-model.json (or FEED_MODEL_PATH)
• Data sources:
  • investorInterests (LIKED/COMMITTED) for supervised positives
  • sampled negatives from unseen products per investor
  • optional behavioral signal enrichment from feedEvents

The runtime ranker in lib/feed/ranking.ts loads model weights via lib/feed/model.ts. If no model artifact is found, it falls back to deterministic heuristic scoring.

The feed now records investor interaction events (impressions, watch milestones, product opens, like/unlike, message click, commit) via POST /api/feed/events into DynamoDB table DYNAMODB_TABLE_FEED_EVENTS.

• app/ - App Router pages and API routes
• components/ - UI and feature components
• lib/db/ - DynamoDB client, table config, repository layer
• scripts/ - local utility/setup scripts
• prisma/ - Prisma schema/migrations/seed artifacts

Current application runtime reads/writes through lib/db/repository.ts (DynamoDB). Prisma artifacts remain in the repository for schema management/legacy workflows and are not the primary runtime datastore for app features.