AJ Barnett
2026 · Product Engineer
Ask A.J. AI Portfolio Agent
Designed and engineered a recruiter-facing AI agent with streamed responses, server-side OpenAI integration, request safeguards, fallback models, cached answers, and graceful degradation.
- Challenge
- The portfolio needed to show AI product engineering as a working system, not just a claim in copy.
- Role
- Product Engineer
- Approach
- Kept provider calls, secrets, validation, retries, and fallback behavior server-side.
- Streamed application-owned NDJSON events so the browser never depends on provider-specific event names.
- Added request validation, origin checks, local rate limiting, duplicate request protection, token caps, and graceful fallback answers.
- Structured the knowledge source so future retrieval can build on curated portfolio facts.
- Impact
- Ask A.J. gives recruiters a direct way to understand AJ's work while demonstrating production-minded AI interface design.
- Lessons
- “An AI product is the whole system around the model: context, trust boundaries, latency, failure behavior, and cost.”
Stack
