BeIntern - AI Based Engine
Maintain your curiosity,and we provide opportunity
Created on 1st November 2025
•
BeIntern - AI Based Engine
Maintain your curiosity,and we provide opportunity
The problem BeIntern - AI Based Engine solves
CareerGuide AI Chatbot
Skills aur goals ko samajhkar short actionable plan deta hai.
“Build X” bolne par project breakdown (scope → setup → MVP → tests → deploy).
OpenAI key ho to AI answers; nahi ho to local rule-based guidance.
Code blocks render hote hain, toh instructions/code clearly dikhta hai.
Unified Mentor form (Ask + Project Submission)
Ek hi form se query + project files/folder upload.
File limits, error messages, aur Google sign‑in gate for safe uploads.
Auto‑review heuristics: repo structure, scripts, tests, tech hints → quick feedback.
Isse students ko jaldi actionable suggestions milti hain.
Curated Data Sections
Internships, Courses, DSA content ko clean UI mein present; search se jaldi filter.
Blog cards hover feedback se scanning fast hoti hai.
Government Schemes Ticker
PM Internship Scheme aur state portals ke links ek animated ribbon me — trusted, official resources tak seedha access.
Performance + Accessibility
Heavy 3D ko lazy-load/fallback approach (prefers-reduced-motion respected).
Images lazy-load (Mentor/Blogs) → faster loads, low-end devices friendly.
Secrets (API keys) .env.local me — privacy/security.
Challenges we ran into
Form merge + validation conflicts
Issue: Merging “Ask” and “Project submission” created duplicated state and race conditions on submit.
Fix: Unified a single submit handler, normalized state, and added clear client-side guards for file count/size before any network call.
Firebase auth gating
Issue: Anonymous auth allowed uploads without identity; Firestore rules would eventually fail.
Fix: Removed anonymous flow and enforced Google sign‑in before uploads; surfaced friendly errors.
File upload performance and limits
Issue: Large folders caused sluggish UI and storage errors.
Fix: Client‑side checks (max files, total bytes, per‑file size). Batched uploads and added explicit error messages.
Automated review heuristics
Issue: Early heuristics were too generic and missed language‑specific structures.
Fix: Added checks for src/, package scripts, pytest.ini, basic model markers, and README signals to tailor feedback.
3D hero lazy‑loading
Issue: Spline iframe hurt initial load and janked on low‑end devices.
Fix: IntersectionObserver to only load when visible. Added prefers-reduced-motion fallback gradient.
Tilt/parallax accessibility
Issue: Hover transforms felt heavy and caused motion discomfort for some users.
Fix: Wrapped animations with motion-safe: classes and provided reduced-motion fallback behaviors.
Vite env handling
Issue: Missing VITE_ prefix and undefined envs caused build/runtime surprises.
Fix: Standardized on VITE_*, added guards (e.g., don’t render iframe if URL missing), and documented .env.local.
Chatbot code rendering
Issue: Messages including code weren’t readable; triple‑backticks appeared as plain text.
Fix: Added a tiny renderer for … and inline code, rendering monospace blocks with horizontal scroll.
Voice/AI fallbacks
Issue: Web Speech API not available in all browsers; OpenAI key may be absent.
Fix: Feature-detected SpeechRecognition; if VITE_OPENAI_API_KEY missing, chatbot uses rule‑based replies.
Image loading and layout shift
Issue: Mentor/blog images caused CLS and bandwidth spikes.
Fix: loading="lazy" and fixed container sizes/rounded wrappers for stability.
Git hygiene and secrets
Issue: Risk of committing .env.local and build artifacts.
Fix: Expanded .gitignore to exclude envs, node_modules, dist, and api/target.
Tracks Applied (1)
Ethereum Track
ETHIndia
Technologies used