I’m Ereg Erig, a Python-focused software engineer and CS student aiming for AI engineering and agent-development roles. I build clean backend APIs, RAG workflows, tool-calling systems, eval-ready product demos, dashboards, and full-stack tools that are easy to use and maintain.
AI Command Center leads this set because it shows agent architecture, RAG, tool policies, guardrails, and eval thinking. The other projects show the backend and product instincts behind useful software.
A product-style AI assistant pipeline with prompt routing, RAG context packs, tool-call policies, guardrails, prompt package generation, eval rubrics, and human-in-the-loop execution.
An AI-assisted weekly decision screen for coursework: Canvas sync, rubric parsing, responsible discussion drafting, assignment triage, study calendar scenarios, and explainable task ranking.
A local-first budget tracker with paid toggles, income modeling, what-if scenarios, JSON export, transparent formulas, and an AI review concept for plain-English spending explanations.
An interactive study workspace with lessons, IOS-style command practice, subnet drills, quizzes, bookmarks, saved progress, Markdown notes, a timed mock exam, and an AI tutor roadmap.
I’m strongest where backend, data, UI, and AI workflow design meet: making the system understandable, testable, useful, and safe for the person opening it.
Designing FastAPI and Django services, REST endpoints, auth-aware flows, background jobs, and clean integration boundaries.
Turning messy real-world tasks into schemas, state, validation rules, PostgreSQL tables, and useful domain objects.
Building React, HTMX, and responsive interfaces that make workflows obvious without overloading the page.
Designing LLM features with intent routing, retrieval, tool calling, guardrails, evaluation checks, and clear human approval boundaries.
I like the parts of the stack where things either work or they don’t: data models, API contracts, AI workflow boundaries, state transitions, integration edges, and product details that make a tool feel reliable.
My favorite projects start with a real annoyance — school planning, budget tracking, AI assistants, study prep — and end as something usable enough to click, test, and improve.
If you’re hiring, collaborating, or curious about one of the projects, email is the fastest place to start. I read everything.