Technical Capabilities

Search and Information Retrieval

Search is the core of our consumer products. We build and operate:

• OpenSearch clusters powering full-text search over millions of documents, including hybrid retrieval that combines lexical relevance with kNN vector similarity
• Apache Solr deployments for structured search and autosuggest
• Relevance engineering: query rewriting and expansion, stemming and text normalization, geospatial filtering, and autocomplete indexes
• Custom geocoding and places search built on search engine infrastructure

Applied AI and LLM Systems

Our platforms use large language models in production, with cost and reliability controls built in:

• Retrieval-augmented generation (RAG) pipelines that ground LLM answers in our own search indexes and vector stores
• Vector search with embedding models, using both dedicated vector databases (Qdrant) and OpenSearch kNN indexes
• LLM integrations with the Anthropic Claude API alongside self-hosted open-weight models served through Ollama
• Natural language query parsing, intent classification, and conversational search sessions
• LLM-assisted data enrichment at scale: classification, attribute extraction, and content generation across large catalogs
• Response caching, relevance gating, and graceful fallback to conventional search when AI components are unavailable

AI Connectivity and MCP

Our products are available to AI assistants as first-class clients:

• Model Context Protocol (MCP) servers that expose product functionality as tools to claude.ai, ChatGPT, and other agent platforms
• OAuth2 authorization flows for AI connector authentication
• Both streamable HTTP and stdio MCP transports, with API key authentication for machine clients

Application Development

• PHP applications on Laravel with Livewire, Jetstream, and real-time WebSocket features
• Python services with FastAPI, SQLAlchemy, and Pydantic
• TypeScript frontends with React and Next.js, styled with Tailwind CSS
• Native iOS development in Swift and SwiftUI backed by token-authenticated REST APIs
• Public and internal REST APIs with versioning, quotas, and feature flags

Data Engineering

• Web-scale crawling and ingestion built on Apache Nutch, with Kafka streaming pipelines feeding document stores and search indexes
• PostgreSQL as the primary relational store, MongoDB for document data, and Redis for caching and queues
• Python ETL for cleaning, deduplication, enrichment, and reindexing across millions of records
• Feed ingestion (RSS/Atom), link metadata extraction, and third-party API integration (Meta Graph API, Mapbox, Expedia, Stripe, Mailgun, Resend)

Infrastructure and Operations

• Linux production environments managed with nginx, systemd, supervisor, and Docker
• CI/CD with GitHub Actions: automated tests, atomic symlinked releases, health checks, and one-command rollback
• Cloudflare for CDN, image transformation, and bot protection; AWS services (S3, SQS, SES) where they fit
• Automated backups, pipeline monitoring, and uptime instrumentation

Security and Commerce

• Authentication with session, token (Sanctum), and OAuth2 (Passport) schemes, including two-factor authentication
• Role-based access control, hashed API key management, JWT services, CSRF protection, and hardened security headers
• Subscription billing with Stripe, including webhooks, grace periods, and quota enforcement

Contact

Questions about our platforms or technology can be directed to: