SELECTED WORK/Smart Skill Hub (SSH)
AI & Full-Stack Systems·2026·Production Ready & Published

Smart Skill Hub (SSH)

Autonomous Developer Intelligence & Adaptive Career Platform

Smart Skill Hub (SSH) is an end-to-end developer intelligence and technical career acceleration platform engineered as a final-year engineering capstone project and published in IRE Journals (ISSN: 2456-8880). It unifies isolated coding assessment, automated ATS resume optimization, GitHub repository code mining, and interactive video curriculum into a closed-loop adaptive learning system.

Smart Skill Hub (SSH)
>94%ATS Match Accuracy
<200msAI Inference Latency
128MBContainer Sandbox RAM
IRE JournalsPeer-Reviewed
The Problem Statement

Aspiring software engineers experience a fragmented learning journey: practice algorithms in isolated silos, claim unverified skills on static resumes, watch distracting video tutorials, and leave authentic GitHub commits unanalyzed by recruiters.

The Engineering Solution

Smart Skill Hub establishes a continuous verification loop: mining live GitHub repos for architectural patterns, running coding assessments in isolated Docker containers, benchmarking user skills against real industry role taxonomies, and generating quantifiable ATS-tailored resumes.

TECHNICAL ARCHITECTURE

Subsystems & Architectural Decisions

FRONTEND LAYER

React 18 SPA built with Vite, TypeScript, Tailwind CSS, Monaco Code Editor, and TanStack React Query for aggressive client-side caching.

BACKEND LAYER

Express 5 REST API modularized across 15 domain modules, powered by Node.js, Helmet security policies, and rate-limiting.

DATABASE LAYER

MongoDB Atlas managed via Mongoose 8 with strict user document ownership isolation and indexed schema lookups.

SANDBOXING LAYER

Dockerode daemon orchestration running a dedicated `ssh-javascript-sandbox` container with 128MB RAM, 0.5 CPU cap, non-root UID execution, and 5-second wall-clock timeouts.

AIENGINE LAYER

Groq Cloud ultra-fast inference running `openai/gpt-oss-120b` and `openai/gpt-oss-20b` with automatic failover to a deterministic local rule/NLP engine.

Implemented Features & Capabilities

01

Smart Mentor: Context-aware AI coaching streaming personalized advice based on candidate skill gaps and GitHub activity.

02

Developer Intelligence & GitHub Mining: Mines public GitHub commits and repository dependencies to authenticate real development depth.

03

Canonical Skill Profile & Cryptographic Badges: Normalizes competencies and issues verifiable credentials with unique public UUIDs.

04

Dockerized Sandboxed Code Runner: Executes JavaScript/TypeScript algorithms safely against hidden and visible test cases.

05

EduTube Curated Learning System: Distraction-free educational video player with interactive quiz checkpoints embedded at video timestamps.

06

Interactive A4 Resume Studio: Real-time editable A4 canvas with automated ATS keyword tailoring and client-side vector PDF generation.

07

Student Study Circles: Socket.IO real-time channels secured with `scrypt` password hashing and constant-time token comparison.

Key Engineering Challenges Overcome

›Preventing Arbitrary Code Execution: Addressed by isolating submissions in non-root Docker containers with strict cgroup memory and CPU limits and disabled networking.
›Zero Downtime on AI Outages: Engineered a 3-tier fallback architecture: primary Groq model -> secondary lightweight model -> deterministic local NLP regex engine.
›ATS Parsing Accuracy: Integrated multi-format OCR and text extraction supporting PDF, DOCX, and raw text with high keyword precision.

Verified Technologies Used

React 18TypeScriptNode.jsExpress 5MongoDBGroq AIDockerodeSocket.IOMonaco EditorTailwind CSS