Aadit Agrawal

i just like to create cool stuff.

EXPERIENCE

Lead Research Engineer @ CASIR, School of Computing, MAHE

MAY 2025-PRES
  • Engineered an LLM-powered industrial protocol translator parsing IEC 61131-3 to IEC 61499, cutting PLC migration timelines for external manufacturing clients.
  • Architected a massive on-premise production cluster running heavy AI workloads, load-balancing traffic across varied architectures (ARM64 Apple Silicon and CUDA Blackwells).
  • Deployed an edge-compute computer vision pipeline onto NVIDIA Jetson hardware flying on DJI drones, sustaining 30 fps inference directly on the embedded board via RTMP.
  • Built an end-to-end full-stack agentic RAG pipeline processing 2000+ structured layouts via custom OCR tools.

Platform Engineer @ Rossmann New Media 2 Inc.

JAN 2026—PRES
  • Led a full-stack migration from legacy WordPress to a modern React/Next.js ecosystem.
  • Redesigned the UI for true cross-device responsiveness, improving core web vitals and overall LCP by 60%+.
  • Executed a technical SEO overhaul by analyzing and mapping data recovery search patterns, moving keyword rankings into the top 5.

DevOps Engineer @ Rasen AI (formerly Pixa AI), Bengaluru

DEC 2025
  • Focused purely on infrastructure-as-code and reliability, provisioning varied compute nodes (H100/H200/B200) tailored for heavy TTS/STT fine-tuning operations.
  • Designed and implemented robust automated CI/CD pipelines guaranteeing 99.5% deployment success across consecutive API releases.

Founding President @ PRISM - Computing & AI Club, MAHE

JAN 2026—PRES
  • Founded and manage an interdisciplinary collective of 150+ members focused on pushing the envelope across Web Development, embedded IoT, DevOps and Applied AI.
  • Act as a technical bridge by serving on MAHE's Academic Advisory Board, helping adapt curricula to modern tech stacks.

Infrastructure Lead @ MIST (Cybersecurity Club)

OCT 2023-JUN 2025
  • Architected a resilient, dynamic CTF web platform supporting 300+ concurrent players, sustaining 100% uptime across 3 high-load events.
  • Optimized cloud configurations and resource limits, reducing operational overhead by 80%.