Syam Sundar Nallamekala

Projects

Work I contributed to rather than owned, stated as such.

A senior design team and an internship API team. Both shipped, neither was mine alone, and each card says which part was. Separating them from the featured six is not modesty — it is so that the six read as unambiguous.

  • Senior design project, University of Toledo

    SmartPark — AI parking space detection

    Edge-AI vision system: 4MP cameras to a Raspberry Pi 4 running TensorFlow Lite and MobileNet SSD v2, classifying each space as occupied or vacant on a live colour-coded map. No per-spot sensors, no cloud round-trips, no personal data.

    Team of fourThe data labelling and segmentation pipeline: I started on the PKLot benchmark at 90%, hit a wall with overlapping bounding boxes, built a custom dataset from a miniature lot, pivoted to segmentation labelling in Roboflow, and retrained to 99%. Also the market research and concept decision matrix, and the system architecture and privacy framework.

    • TensorFlow Lite
    • Edge AI
    • Roboflow
    • FastAPI
  • Kalasalingam Academy of Research and Education, 2022

    FlowerBot — flower-plucking robotic arm

    Rose harvesting is manual, painful and impossible after dark: thorn injuries, sustained bending, no visibility at night. A PIC-controlled arm with night-vision cameras distinguishes flowers from leaves and bushes, detects obstacles, and slices and holds the flower in one motion, wet or dry. Classification and variety identification used SVMs, k-nearest neighbours, random forests and CNNs, and combinations of them.

    Undergraduate team of sixOne of six authors on the paper and a listed inventor on the granted patent. The work is published and patented, which is why the paper and the patent sit on /research rather than here — this card is the project, not the credential.

    • Robotic vision
    • CNN
    • SVM
    • Embedded

    Paper and patent on Research

  • Park Place Technologies

    Security Discovery Tool API

    Heterogeneous device-discovery feeds ingested into one centralised system. 20% improvement in data accuracy, and a contribution to 99.5% system uptime.

    API team intern, May – August 2024I wrote the Node.js and Express data connectors and the normalisation and preprocessing pipeline that fed them into the central system, and the error handling that kept a malformed feed from taking the ingest down.

    • Node.js
    • Express
    • REST