Profile picture

An AI Systems Engineer with more than two years of experience creating scalable backend architectures and computer vision systems that work in real time. Specializes in improving machine learning models for use in production (ONNX, PyTorch) and creating microservices that can handle a lot of requests at once to boost performance.

Skills
Backend & Concurrency: — Python, FastAPI, Django, SQL, WebSockets, Celery, Asyncio and Multithreading.
AI & Computer Vision: — PyTorch, TensorFlow, OpenCV, MediaPipe, InsightFace, ONNXRuntime, Transformers, XGBoost.
DevOps & Infrastructure: — Docker Swarm, Linux (Ubuntu), FFmpeg, Cloudflare, Vercel, Koyeb, PostgreSQL (Neon).
Work Experience
Freelance, AI & Backend Developer
  • •Creating a Diffusion pipeline that works best for processing webcams in real time and can be used with a wide range of client applications through an API.
  • •Using PyQt6 and PyOpenGL to render 3D models with accurate depth tracking and occlusion, I made a desktop app that lets you try on jewelry in real time.
    • 12/2025 – PresentRemote
  • •Used FastAPI for the backend, Next.js for the frontend, and a Neon PostgreSQL database to build and host a full-stack e-commerce platform on Vercel and Koyeb.
    • Autobits Labs⁠, AI Systems Engineer
  • •By changing an old CPU-based face recognition system to a custom InsightFace and ONNXRuntime GPU pipeline, we cut inference latency in half.
  • •Built a high-concurrency FastAPI backend with WebSockets that can handle multiple live MJPEG streams from multiple RTSP cameras at once without stopping the main event loop.
    • 08/2024 – 11/2025Rajkot, India
  • •By making a hybrid Transformer and XGBoost pipeline for spectral analysis of chili powder, we were able to predict the amount of Capsaicin with 85% accuracy (R2).
  • •Made custom .deb packages for installing Docker completely offline and set up Samba NAS and dual-mode networking on Raspberry Pi for environments that are air-gapped.
    • Nijitek Pvt. Ltd., Software Engineer (Backend)
  • •We built an automated Django platform that made personalized children's media on the fly, which cut down on the time it took to make content by 90%.
  • •Configured Celery workers to process peak order queues without any problems, which improved backend stability and automated product generation.
    • 02/2024 – 07/2024Ahmedabad, India
  • •Used OpenCV and FFmpeg to make a dynamic media pipeline that automatically puts together images and audio into video slideshows.
    • Projects
    Real-Time Indian Sign Language (ISL) Recognition
  • •Created a gesture recognition system with low latency using MediaPipe.
  • •Created a multithreaded Text-to-Speech engine that gives instant audio feedback without stopping the video input stream.
    • 08/2025 – 09/2025
    Education/Training
    Training Course, Pyspiders

    Python Full-Stack Development Course.

    03/2023 – 09/2023Bengaluru, India
    B.Tech., Atmiya University

    (73.54 % | CGPA: 7.82) Electrical Engineering.

    08/2018 – 03/2022Rajkot, India