const vikash: Engineer = {
name : "Vikash Kumar",
role : "AI-Native Full Stack Engineer - SDE-1 Aspirant",
education : "B.Tech CSE-IoT @GEC Siwan, Bihar | Oct 2022 – May 2026 | CGPA: 7.27",
available : true,
phone : "+91 736-288-5182",
targeting : ["SDE-1", "Full Stack", "React Native", "Backend", "AI-Integration"],
production : {
company : "Sewacity Technologies Pvt Ltd",
users : "10,000+",
transactions : "2,000+ / month",
playStore : "https://play.google.com/store/apps/details?id=com.sewacity.sewacity",
reliability : "+40% payment success via Razorpay webhook orchestration",
growth : "+36% order growth via latency-optimized REST APIs",
security : "-60% unauthorized access via JWT/RBAC + rate limiting + DDoS",
failures : "-30% payment failures via retry logic + reconciliation",
},
stack : {
languages : ["TypeScript", "JavaScript", "C++", "HTML", "CSS", "SQL", "NoSQL"],
frontend : ["React", "Next.js", "Tailwind CSS", "shadcn/ui", "Redux", "Zustand"],
backend : ["Node.js", "Express.js", "REST APIs", "WebSockets (Socket.io)", "JWT/RBAC"],
mobile : ["React Native", "Expo EAS", "Firebase OTP", "AAB/IPA signing", "TestFlight"],
database : ["MongoDB", "Redis", "MySQL", "PostgreSQL", "Supabase", "GridFS"],
cloud : ["AWS EC2/S3/Route53", "Docker", "Kubernetes", "NGINX", "PM2", "CI/CD"],
ai : ["MCP Protocol", "LLM SDK Integration", "Claude AI", "ChatGPT", "Copilot"],
},
dsa : "600+ problems | C++ | LeetCode + GeeksforGeeks + Codeforces",
hackathons : ["IIT BHU Shilp'24 - 2nd Place (50+ teams)", "TechKshitiz GEC Siwan - 1st Place"],
leadership : "Technical Team Lead @TechKshitiz, GEC Siwan - 20 months",
vision : "SDE-1 → Senior Engineer → CTO / Technical Co-Founder",
};Not a chatbot wrapper. Real architectural integration of AI and LLM reasoning into production systems via MCP Protocol - the same pattern powering Claude, Cursor, and next-gen AI tooling.
User Input → LLM (Claude/GPT) → MCP Protocol → Tools / APIs / DB / IoT → Result
│ │ │ │ │
Natural Reasoning JSON-RPC 2.0 MongoDB • Redis Structured
language over live Tool contracts RFID • REST output back
prompt context (Host→Client→Server) Sensors • GridFS to product UI
My AI Integration Stack
How I use each tool in my actual workflow:
- Claude AI - architecture decisions, writing technical documentation, code review, system design walkthroughs, and MCP server implementation. I don't use it to write code blindly - I use it to think through design tradeoffs before touching the keyboard.
- ChatGPT - quick syntax fixes, exploring multiple approaches to the same problem, rubber-duck debugging on unfamiliar APIs, generating test cases.
- GitHub Copilot - in-editor code completion, reducing boilerplate in repetitive patterns (route handlers, schema definitions), and surfacing API signatures mid-flow.
- Antigravity AI Editor - AI-native editor for faster development loops with embedded LLM assistance.
MCP in production (Academic Management System):
// MCP Server: academic-intelligence
// Architecture: Host (Claude) → MCP Client → MCP Server → MongoDB + RFID sensors
// Transport: JSON-RPC 2.0 over stdio
const tools = [
{
name: "get_student_recommendations",
// LLM receives live student context → reasons → returns structured suggestions
// Zero prompt leakage. Typed contracts. Auditable tool calls.
},
{
name: "analyze_performance_patterns",
// Connects LLM reasoning to MongoDB aggregation pipelines directly
}
];
// Result: AI-powered personalized resource recommendations running in production
// User → LLM → MCP → MongoDB + RFID → Personalized response| Metric | Result | How |
|---|---|---|
| Users Served | 10,000+ | Web platform + React Native app (Play Store) |
| Monthly Transactions | 2,000+ | Razorpay payment pipeline |
| Payment Reliability | +40% | Webhook orchestration + retry logic + idempotency |
| Order Growth | +36% | Latency-optimized REST APIs + geo-aware backend |
| Unauthorized Access | -60% | JWT/RBAC + rate limiting + DDoS safeguards |
| Payment Failures | -30% | Failure reconciliation layer + dead-letter logging |
| Transaction Integrity | 100% | Server-side Razorpay HMAC-SHA256 verification |
| GitHub Contributions | 2,132 | Last 12 months - consistent daily engineering |
Languages
Frontend
Backend
Database
Mobile (React Native)
Cloud, DevOps & Infrastructure
Developer & Productivity Tools
Figma: I read and implement designs, convert component specs to production-ready code. Design creation is not my primary strength - shipping designs as pixel-accurate components is.
|
Sewacity Mobile App - iOS + Android
Production React Native app - 10K+ users, 2K+ monthly orders. Redux state, Firebase OTP auth, server-side Razorpay verification, real-time order tracking, geo-location. Expo EAS builds with AAB signing (Play Store) and IPA distribution (Apple Dev Console + TestFlight). |
SafeExec - Enterprise RCE Platform
Secure open-source Remote Code Execution platform. Every execution runs in an isolated Docker container - no shared state. Redis job queue for scalable throughput. Multi-language support (Python, JS, Java, C++, C#, Go). Swagger UI docs. Built for exam platforms, hiring tools, online judges. |
|
Contest & Event Management Platform
Built for TechKshitiz, GEC Siwan - ran a live contest for 300+ students simultaneously in production. Automated seating assignment, participant verification, integrated code editor, real-time test workflows. Designed to scale to institutional exam bodies. |
Academic & Library Management System (AI + IoT)
Campus management with RFID automation, MCP-based LLM recommendation engine, layered RBAC, coursework workflows, scheduling, and digital repository. Real MCP server implementation connecting LLM reasoning to live MongoDB + IoT sensor data. |
|
MongoDB GridFS Bucket System
Open-source file storage API on MongoDB GridFS. Store, stream, and manage large files via a clean REST interface. Documented and production-ready. Solves the S3-overkill problem for small-to-medium file management. |
Portfolio v3 - Production SPA
|
Problem → Decompose → Data Model → API Contract → Security Layer → Fault Paths → Ship
Real example @Sewacity - Razorpay Webhook Reliability
──────────────────────────────────────────────────────
Problem : Payment status inconsistency - client-side verification unreliable
Root : No idempotency, no retry on failures, no reconciliation job
Solution : Server-side HMAC-SHA256 signature verification
+ Idempotency key per webhook event
+ Redis retry queue with exponential backoff
+ Cron-based failure reconciliation job
+ Dead-letter log for manual audit trail
Result : +40% transaction success, -30% failures, 100% integrity
|
|
|
|
50+ competing teams - full-stack solution under time pressure |
GEC Siwan - led and won the hackathon |
TechKshitiz org, GEC Siwan |
|
|
|
DNS, SSL, production server management |
|
Funded Startups (Seed / Series A) Full stack from mobile to infra. Built for 10K+ users as a solo contributor. I don't wait for tickets - I find bottlenecks, scope the fix, and ship. Built for early-stage velocity where every engineer counts. |
MNCs and Product Companies Security-hardened APIs, event-driven design, fault-tolerant webhook systems. Docker, AWS, CI/CD, NGINX in production. Strong DSA foundation (600+ problems). Code that survives proper code review. |
AI-First Products I understand MCP protocol architecture end-to-end - not just API calls. I've shipped LLM-driven features in production with real data contracts and typed tool schemas. No brittle prompt hacks. |
Open for → SDE-1 / Full Stack / React Native / Backend / AI-Integration roles
Building → Razorpay Webhook Debugger SaaS (MVP scoping)
Integrating → Meta Ads FBSDK + Conversions API (server-side, production)
Learning → MCP tooling, AI-connected developer utilities
DSA → Medium → Hard (Graphs, DP, System Design)
Vision → SDE-1 → Senior Engineer → CTO / Technical Co-Founder
Open to SDE-1, Full Stack, React Native, and Backend roles at startups and product companies.
Vikash Kumar - Built with intent. Every commit ships toward production.