learning never hits exit, only run.
|
ROLE |
3rd year,Btech, Electronics and Computer Science Engg. |
Equally comfortable writing the firmware that reads a sensor and the pipeline that decides what to do with what it read. Recent work sits at that seam -safety research on software-defined vehicles, fault injection in simulation, ECU and kill-switch circuits on an actual kart, and AI systems that never touch hardware at all. Same instinct either way: find where the signal is lying to you, then fix it.
| TIME | PROCESS | TAGS |
|---|---|---|
2026 · MAY–JUL |
SDV & AI Automotive Safety Research Intern -IIT Indore | SDV · SOTIF · Rust · STM32F407 |
2026 · FEB–PRESENT |
Electrical & Firmware -Team Full Throttle (GKDC Kart) | C/C++ · ECU integration · kill-switch design |
2026 · MAY–JUN |
Research Intern -Open Knowledge Initiatives × IIIT-H | SPARQL · Wikidata · OCR pipelines |
2025 · JUL–PRESENT |
Committee Member, Sponsorship & Mgmt -Aaruush | corporate pitching · MoUs |
2025 · DEC |
Industrial Trainee -Pi-Tech India (MSME Certified) | Arduino · ESP32 · UART/I2C/SPI |
Data centres get sited and scheduled around latency and cost -almost never around water. NeerAI adds regional water stress as a real scheduling variable, not a footnote, and routes compute accordingly.
what it does -profiles workloads, predicts resource pressure, and runs a multi-objective optimizer (J = αC + βE + γWs + δL) across compute, energy, and water stress before a job ever gets placed.
what's built so far -the optimizer, a Python digital-twin harness for simulating infrastructure telemetry, and a routing layer that turns four competing variables into one decision. Still actively evolving.
Python · XGBoost · optimization · digital twins · telemetry
|
Full-stack platform delivering personalized tech content through a recommendation engine, with JWT + Google OAuth auth and a MySQL schema built around a user-interest graph.
|
|
AI-vision + sensor fusion system for automated railway track defect monitoring. ESP32 nodes process ballast health and obstacle telemetry in real time; loco-aware braking logic computes emergency stopping distances on hazard detection.
|
Converts unstructured news into a navigable knowledge graph -custom adjacency-list structures, binary-heap priority queues, and Union-Find powering Dijkstra's and Kruskal's traversal modes, with LLM-based keyword extraction on top.
|
|
ESP32-based closed-loop controller processing real-time soil-moisture telemetry to trigger automated irrigation cycles. Implemented software hysteresis to kill relay chatter and cut power draw.
|
Cross-platform app that treats skill growth like a dependency graph. LLM APIs generate adaptive roadmaps and missions with stateful context across sessions; Firebase handles real-time sync and auth.
|
|
Off-chain Python engine using Merkle-tree inclusion proofs to verify ad-impression records, with Solidity contracts that process submission hashes on-chain and auto-flag disputed activity.
|
|
FIELD NOTES
[2025] SIH National Finalist -RailSentinals, AI-vision track hazard monitoring system
[2026] IIT Indore -SDV / SOTIF safety research
[2026] OKI × IIIT-H -performance stipend, cohort of ~140
[——] National Spelling Bee -3rd place
| LANGUAGES | Python · C · C++ · JavaScript (ES6+) · SQL · Solidity · Rust (foundations) |
| WEB & BACKEND | React · Node.js · Express · REST APIs · JWT · Google OAuth · HTML/CSS |
| AI & AUTOMOTIVE | ML Fundamentals · LLM APIs · SOTIF · SDV Architectures · CARLA · LGSVL · NVIDIA DRIVE Sim · IPG CarMaker |
| EMBEDDED & HARDWARE | STM32F407 · ESP32 · Arduino · CAN / UART / SPI / I2C · Digital Twins · OTA Updates |
| TOOLS & INFRA | Git · MySQL · Firebase · Linux · Docker (learning) · Microservices (learning) |