Electronic Portfolio — 2026 Build

A hub-and-spokes portfolio covering: Software Engineering (including AI-integrated projects like RAG pipelines), Competitive Programming, Machine Learning, and Embedded Systems & Robotics.

Tech Stack

Frontend

• Next.js 15 (App Router) + React 19 + TypeScript
• Tailwind CSS 3 — editorial off-white baseline used on every spoke
• Arwes (@arwes/react) — sci-fi HUD framework, hub page only (dark cyberpunk exception, see CLAUDE.md)
• Framer Motion / Motion — per-spoke background animation (Minecraft day/night cycle, CP terminal loop, ML robot arm, space float)
• MDX (@next/mdx, @mdx-js/react) — case study write-ups in content/case-studies/

Backend (stubs)

• FastAPI + Pydantic (Python) — services/ml-service/, services/agent-service/

Live data integrations

• Codeforces API — CP spoke rating/contest history
• LeetCode's internal GraphQL endpoint (no official public API) — CP spoke solved-count/contest stats

Deployment

• Vercel — frontend hub + all spokes
• Root Directory set to frontend; see frontend/.npmrc (legacy-peer-deps=true, required by Arwes's current prerelease against React 19)

Structure

app/                    # Next.js hub (App Router)
  page.tsx              # Home — links to all four spokes
  spokes/                # One placeholder page per core, replace with real content
    swe/
    cp/
    ml/
    embedded-robotics/
  api/cp-stats/         # Route that fetches live Codeforces stats
content/case-studies/    # MDX write-ups per project (see example.mdx)
services/
  ml-service/           # FastAPI stub — deploy ML demos here
  agent-service/        # FastAPI stub — deploy the RAG/agent demo here
embedded/
  firmware/             # Microcontroller code
  docs/                 # Wiring diagrams, build logs, photos
.github/workflows/       # CI + a cron stub to refresh CP stats

Getting started

npm install
npm run dev

Open http://localhost:3000.

For the Python services:

cd services/ml-service        # or agent-service
pip install fastapi uvicorn
uvicorn main:app --reload

Deploying

• Hub: https://e-portfolio-herrdelta83s-projects.vercel.app
• ML/Agent services: deploy each to Hugging Face Spaces or Railway; point the hub's spoke pages at the deployed URL (iframe or fetch).

Overall Projects

Universal (every repo)

• ✅ Live demo, gif, or screenshot above the fold — not buried in the code
• ✅ Commit history showing real iteration, not a single dump
• ✅ Tests included, even minimal
• ✅ Scoped to do one thing well

SWE

• ✅ Architecture decisions explained with tradeoffs, not just a tech-stack badge wall
• ✅ CI/CD pipeline (green badge = "ships," not "sits in a folder")
• ✅ Real users where applicable (e.g. SEITC platform — used by N students)

Competitive Programming

• ✅ Algorithmic knowledge applied to a real problem (not a leetcode-solutions dump)
• ✅ Live ratings & standings tracked via Codeforces / ICPC badges above

ML / Data Science

• ✅ Evaluation shown, not just training — metrics table, baseline comparison, error analysis
• ✅ Data handling made explicit (cleaning, leakage checks, splits)
• ✅ A real plot (loss curve, confusion matrix, forecast vs. actual) over a paragraph of claims

Embedded Systems & Robotics

• ✅ Video of the physical system working — non-negotiable in this domain
• ✅ Evidence of a closed feedback loop (PID tuning graph, SLAM map output, control loop diagram) — not just "wired a sensor"
• ✅ Documented failure modes — what didn't work, and why
• ✅ Wiring diagram / schematic included