---

>  NAME         →  Aariyan Sunu   >  LOCATION     →  Kerala, India   >  DEGREE       →  BTech CSE · 7th Sem · MITS · KTU   >  ROLE         →  Backend Eng + AI Systems Builder   >  MINDSET      →  "I think in systems, not websites"   >  FUN FACT     →  Debugs 40% faster with lo-fi Currently shipping Domain Depth Backend Systems ████████████████████░░ 90% AI / ML Pipelines ████████████████░░░░░░ 75% Frontend / React █████████████░░░░░░░░░ 65% Hardware / ESP32 ████████░░░░░░░░░░░░░░ 40% DevOps / Infra ███████░░░░░░░░░░░░░░░ 35%

---

Jarvis — Voice AI Assistant      Local-first voice OS replacing the keyboard entirely
Pipeline: Wake Word → Voice Auth → Whisper STT → Cache → Gemini Router → ReAct Agent → ElevenLabs TTS Porcupine wake word + ECAPA-TDNN speaker verification Gemini 1.5 Flash intent router — sub-100ms on 50 cached commands ReAct multi-step agent — up to 5 chained planning steps Mood-based ElevenLabs TTS (happy / focused / tired) Iron Man HUD — radar, arc reactor, voice auth, biometrics panels ESP32 wearable terminal over WiFi via ngrok and Flask on Mac
Smart Emergency Response Simulation      Real-time ambulance routing with live ETA prediction
Simulates an entire emergency dispatch system — live map, moving ambulances, adaptive ETA, and real-time updates for all connected clients simultaneously. React + Leaflet — interactive map with live ambulance movement WebSocket broadcast — state synced across every connected client Online learning model — ETA refines itself with each dispatch Route optimizer — shortest path under dynamic conditions Dispatch dashboard — full metrics view per incident
AI Pothole Detection System      Computer vision microservice for smart city infrastructure
A plug-in CV microservice that classifies road damage severity from images, returning structured results fast enough to embed into any smart city pipeline. MobileNetV2 fine-tuned on road damage dataset FastAPI inference endpoint — sub 200ms per image Returns { severity, confidence, bounding_region } as JSON Modular design — drop into any city monitoring pipeline Batch inference support for bulk image processing
Centralized Log Ingestion Platform      Production-grade observability infrastructure from scratch
End-to-end observability stack — a drop-in npm logger, a central ingestion server, full-text log search, and email alerting. Built for distributed microservice monitoring. Custom Node.js logger — npm-ready drop-in library Ingestion server with full-text search across all services Alert threshold engine — email hooks on error spike detection Horizontal scale design — handles multi-service log streams Filter by service, level, and time range

---

Languages

Backend

Frontend

Databases

AI / ML

Hardware & Tools

---

---

---

Jarvis — Next Modules

ML Roadmap

---

+--------------------------------------------------------------------+
|                                                                    |
|   "I think in systems, not just websites.                          |
|    I build things that solve real problems -- from voice AI        |
|    assistants to emergency routing engines."                       |
|                                                                    |
+--------------------------------------------------------------------+