Jerik David H.

Most clients come with a problem that's already costing them time, money or both. What I do is turn that into a system that runs on its own . I've been building with LLMs since before ChatGPT made it mainstream, which means I've seen how this technology actually behaves at the edges, not just the happy path. That experience changes how I approach constraints, failure handling and output validation in ways that matter in production. Some examples of what I've shipped: - Conversational AI platform for sales and support, full agent architecture with context handling and business logic - SaaS marketplace for website delivery, Stripe payments, dynamic product configuration and purchase flow tracking - Full operations platform for a service business: reservations, client management, quoting and billing - Compliance automation agent: daily scheduled login, MFA handling, data processing and per-location reporting - LangChain-based systems, content classification models and data pipelines for multiple long-term clients I use Claude Code and Cursor as core development tools and integrate LLMs into real workflows where the output has actual consequences. That means as much time on constraints, validation and failure handling as on the feature itself. AI & LLM: Claude API, OpenAI GPT, LangChain, prompt engineering, RAG pipelines, agent architecture, fine-tuning, output validation, Claude Code, Cursor Full-Stack: Python, TypeScript, Next.js, NestJS, React, PostgreSQL, Prisma, REST APIs, Stripe, Supabase Infrastructure & Automation: Docker, AWS, Linux, systemd, CI/CD, GitHub Actions, Playwright, web scraping, scheduled agents If you need someone who can own the technical side end to end and move fast without cutting corners, that's exactly how I work.