Backend and Machine Learning Engineer specializing in production-grade, latency-optimized systems. Currently pursuing a B.E. in Computer Science and Engineering (AI & ML) at Saranathan College of Engineering.

My work focuses on the operational layer of intelligent systems — inference optimization, retrieval architecture, and backend design under real constraints: latency, memory, and cost. This approach has produced measurable outcomes, including a 51% reduction in CPU inference latency on an edge deployment and a sub-14ms client-side retrieval engine.

I approach engineering as a discipline of evidence: every claim is backed by a benchmark, and every system is built to hold up beyond the demo.

Languages

Frontend

Backend & Databases

AI / ML

Cloud & Tools

YOLO-Vision-X — real-time object detection, CPU-only. Swapped the Ultralytics wrapper for a hand-optimized OpenVINO IR pipeline: 270ms → 132ms, 51% cut, no GPU. 80-class tracking with persistent IDs, multi-stream switching that doesn't restart the server, intrusion-zone alerts over SSE. repo

SCE Student Portal — live, multi-role campus platform (with Madhav Padmesh S). Retrieval-grounded answers instead of a static FAQ. In actual use. repo

AnswerFlow — semantic search that never leaves the browser. Zero backend, zero API calls, sub-14ms per query via precomputed vectors + BM25-style weighting + fuzzy correction. repo · demo

Velora — 42-language translator on Groq inference, 111kB bundle, zero production errors. Race-condition guards on rapid concurrent requests. repo · demo

Java Programming Intern · CodeAlpha · Feb 2026 Virtual · MSME Registered · Cert CA/DF1/21283

Three production Java modules — OOP hierarchies, custom exceptions, Collections-based pipelines. Zero critical defects across code review.

building:
  - RCMS — conference management, Java / Spring Boot

exploring:
  - Multi-agent orchestration (Google ADK)
  - Edge deployment on constrained hardware

open_to:
  - Backend / ML systems internship
  - Research internship