Eddie
Ai Systems Engineer

AI / ML Engineering

• Practical AI systems: chat, agents, and copilots with tool use/function calling, RAG, structured outputs, and observability.
  
  

• Models: GPT-4o/mini, o3, Claude 3.5, Llama 3.1—selecting for cost, latency, and capability.
  
  

• Retrieval: Postgres + pgvector, embeddings pipelines, chunking, re-ranking, and safety rails.
  
  

• Evaluation: prompt A/B tests, guardrails, and offline evals for hallucinations, latency, and cost.
  
  

• Vision & speech: image understanding, TTS/STT, and multimodal agent flows where it actually helps.
  
  

• Tooling: plain REST (no lock-in), light LangChain/LlamaIndex when it saves time.
  
  

Web Engineering (Modern Frontend)

• Stack: Next.js 15 (App Router), React 18, TypeScript, Edge APIs, Vercel.
  
  

• UX: clean, keyboard-friendly interfaces; gesture-first audio; streaming UI when beneficial.
  
  

• Content: MDX for long-form writing & case studies; SEO-friendly routes and metadata.
  
  

• Perf: image optimization, route-level code-splitting, and fast TTFB on the Edge.
  
  

• Styling: Tailwind, accessible components, and production-grade polish.
  
  

Backend & APIs

• Node/TypeScript services with clean HTTP contracts, auth, and rate limits.
  
  

• Task runners and queues for background work (webhooks, ingestion, ETL).
  
  

• Observability: logs, metrics, and simple dashboards—know when something breaks before users do.
  
  

Data & Databases

• Relational first: PostgreSQL with Prisma for schema discipline and speed.
  
  

• Vector search: pgvector when retrieval matters; Redis for hot paths & caching.
  
  

• Modeling: schema design, migrations, and query performance tuning.
  
  

Cloud & DevOps

• Deploy: Vercel for web/edge; AWS (Lambda/S3) or Azure when needed.
  
  

• Containers: Docker; Kubernetes only when scale and reliability demand it.
  
  

• CI/CD: GitHub Actions for build, type-check, lint, and Playwright smoke tests.
  
  

• Cost control: budget-aware choices (token usage, caching, cold starts).
  
  

Audio / Music Systems

• AI-assisted music pipelines and Web Audio integrations.
  
  

• Studio skills: recording, mixing, mastering; ffmpeg tooling for media automation.
  
  

• UX: persistent players, auto-pause with HTML5/YouTube detection, gesture-safe playback.
  
  

Blockchain (Selective Use)

• Smart contracts and crypto payments where they fit product goals.
  
  

• Focus on practical automation and verifiable workflows—no hype for hype’s sake.
  
  

IT / Security Foundations

• Systems administration, cloud access controls, backups, and incident basics.
  
  

• Threat-aware defaults: secrets management, least-privilege, audit trails.
  
  

Project Leadership

• Outcome-driven delivery: clear specs, short iterations, tight feedback loops.
  
  

• SOPs & hand-offs: docs that make systems maintainable by someone else later.
  
  

• Stakeholder alignment: translate goals → roadmaps → measurable wins.
  
  

Working Style & Principles

• Simplicity beats complexity. If a simpler solution delivers the same value, we ship that.
  
  

• Open and transparent. Clear code, clear prompts, clear docs.
  
  

• Future-proof but frugal. Choose tools you can own and operate cheaply.
  
  

• User impact first. AI is a feature, not the product—unless it is the product.