Meenakshi Narayan Ramachandran

Address: ***, ******** **********, *** ***, Milpitas, CA, 95035

Email: adzkpe@r.postjobfree.com

Cell: +1-352-***-****

www.linkedin.com/in/meenakshi-narayan-ramachandran A Data enthusiast with over 2 years of professional experience as a Project Manager. Worked extensively in a client facing experience while gathering customer requirements and successfully closed over 10 projects with good client reviews. Contributed to projects that includes coding in Python using Computer Vision Annotation Tool. Furthermore, used YoloV4 model and OpenCV python packages to perform object detection of real time objects using RealSense camera. In addition to my technical skills, my communication and interpersonal skills are excellent. Education:

B.Tech – Electronics and Communication, Amrita School of Engineering, India Aug 2015 – Aug 2019 M.S. - Electrical and computer engineering, University of Florida Aug 2021 – Dec 2022 Skills:

Programming Languages – Python (packages: OpenCV, NumPy, Pandas, TensorFlow Keras, Pytorch); Software – MATLAB, Oracle, MS office, AVR Studio, Arduino, gem5, Visual Studio, PyCharm, Open vino, Cadence; Hardware – Arduino, AT mega board, RealSense Camera; Cloud Platforms – Google cloud, GitHub.; Operating systems – Windows, Linux (Ubuntu), Mac; Web interface – Gradio. Professional Experience:

ICURO, Santa Clara - Computer Vision Engineer April 2023 – Present

• Benchmarking and training of the UI developed for measurement analysis utilizing API request calls.

• Worked on Analysis scripts to create standalone image visualization for the backend data.

• Debugged YOLOV4 and GNN models to get inference measurement and accuracy reports. ICURO, Santa Clara - Computer Vision Intern May 2022 – August 2022

• Measurement analysis of silicon wafer – using python OpenCV and genetic algorithm packages to perform image transformation and measurements.

• Trained YOLOv4 Darknet model for object detection and classification of the datasets for autonomous mobile picking robot and Intel’s Open Vino tool was used to analyze model performance and accuracy.

• Generated the inference graphs using python to accept a real time input from RealSense camera using the trained model weights to perform object detection.

INFOSYS Private Limited, India - Project manager Nov 2019 – July 2021

• Interfacing with an offshore client company- TELSTRA and their customers to accomplish end to end connectivity in real-time by using software’s like Cisco Jabber, IDEAL, AMDOCS, TeCAMS and many more.

• Single handedly managed all the phases of the project from initiation, planning, execution, monitoring, control, billing and finally project closure.

Relevant Projects:

• A Robust Indoor Mapping Algorithm using Multi view stereo and scan matching technique (to acquire an indoor image using LIDAR sensor, extract features by incorporating non-linear MVS algorithm which details the distance of an object surface as a depth information of the scene object. This gives a pixel-to-pixel image representation as a compact 3D point clouds using Visual SFM software. Images was reconstructed using sparse/ dense reconstruction mechanism and ROS package workspace).

• Activity Recognition using Python and TensorFlow Keras (using PyCharm tool and pretrained Inception V3 model to infer activity of the test dataset).

• Smoothing of medical image contours (using PyCharm tool and OpenCV python packages the edge points of discontinuity was detected in a grayscale medical image, the least mean square of the Euclidean distance between two successive edge point is calculated. FFT is calculated for all these points and masked over the original image to obtain the smoothened contour).

• Recognition of handwritten ASCII symbols using CNN (Inception V3 train model was used to learn the different train image sets. This model runs parallel process to implement kernels of different sizes on the extracted features).

• Interactive high humidity detector (ESP32 and HTU21D temperature and humidity sensor were integrated over Arduino IDE to send the humidity data to the AWS server and printed using inbuilt MQTT test client. If the humidity detected was greater than 100, then an email will be sent to the user according to the query written using AWS SNS push services).

• Fingerprint Recognition (The image was acquired and pre-processed. Feature extraction to locate, measure and encode the ridges was done using SIFT OPENCV python package that calculates the Laplacian of Gaussians to find the potential key points. The least square Euclidean distance is calculated to match the key points in the train and test data set for the accuracy score).

• Design and Analysis of an 8X4 SRAM (Using Cadence software memory circuit for 8x4 SRAM was constructed using 45nm technology node. The circuitry, simulation and layout were separately done for SRAM cell, array, decoder, read/ write circuit, and sense amplifier. The individual components were integrated and simulated for a successful LVS, DRC and simulation).

• Single Image Dehazing (Using haze line algorithm where non-local priors are tight clusters of RGB that gets translated into haze lines when there is only atmospheric light across the picture. Code was implemented from https://github.com/ danaberman/non-local- dehazing; and tested for different cases and recorded results)

• Implementation of grof2dot and ScoreP profiling tools (Using graphviz package and python to graphically represent the parallel program call functions graphically).

Contact this candidate