AKHIL PRASANNAN

Experience

• 

  SynergyLabs

  Jun 2018 - Jan 2019

  Deep Learning Engineer

  Designing and developing machine learning and deep learning systems and running tests and experiments.● Successfully​ ​ built an Automatic Number Plate RecognitionSystem (​ ANPR​ ) and achieved an ​ accuracy of 99.7 per cent ​ for an end to end number plate recognition. Currently, the system is being used at Vipul Square, Gurgaon to monitor incoming and outgoing vehicles.● Successfully built and deployed an end to end Facial Recognition now currently being used at Jazari Institute Of Artificial Intelligence, Gurugram for marking of employees.● Performed Data Analysis and inspected metadata of a VideoCompression tool developed at SynergyLabs. Cleaned and visualised the generated data and created report a performance report the client and for tuning the Deep Learning model performance. Show less

• 

  Poseidon-AI

  Jun 2019 - Sept 2019

  Computer Vision Engineer

  Worked on different computer vision problems specific to aquaculture which included Floating feed weight calculation, Fish fIngerling counting from the video feed and fish classification.

• 

  Datakalp

  Oct 2019 - Feb 2020

  Associate Data Scientist

  ● People Detection And Tracking For Retail: ​ Implemented and evaluated the performance of multiplePeople Detection algorithms and added multiple object tracking.● Multiple Object Tracking: ​ Implemented and tested multiple object trackers.Developed custom tracklet prediction and tracklet association logic● A Generalised pipeline for detection, tracking and recognition​ : Implemented a more generalisedpipeline for computer vision problems where multiple object detectors, trackers and classifiers can be swapped for both testing and can be used in a production environment.● Test Suite for End to End Object Detection and Recognition Pipeline​ : Developed a test suite to assess the performance of end to end object detection and recognition pipeline. Show less

• 

  Spyne

  Mar 2020 - Aug 2021

  Computer Vision Engineer

  ● Implemented Entire Backend for Inference: Developed and deployed diverse AI services Background-removal(Saliency Detection)/replacement, Image tagging, Shadow generation on multiple cloud services(Amazon AWS, Google cloud and Microsoft Azure) with focus on performance.● Human Parsing for Automated Image Cropping : Built and deployed a deep learning module using Pytorch ,Opencv for cropping images for E-commerce and deployed on Azure. Usage of this module saved an approximate additional 30% of an editor's time.● Multiple POCs on Object Detection,Segmentation, Image Clustering,2D to 3d generation using NERF. Show less

• 

  Upscalepics

  Nov 2020 - Aug 2021

  Early Team Member
• 

  Quantiphi

  Aug 2021 - Jun 2024

  Large Scale Image Classification models (Computer Vision): Trained and deployed multiple variations of CNNs and Transformer based networks for 2000 plant classification and 10000 animal classification. Achieved 3X faster training with the same model architecture and training hardware configurations.● Multi-node Training: Implemented multi-node training on AWS Sagemaker for custom Deep learning models.● Intent Identification and Entity Extraction(NLP): Implemented intent identification and entity extraction module using both inbuilt aws comprehend and pytorch.● Image Segmentation POC (Computer Vision): Implemented Image Segmentation for detecting the type of underground optical fibers. Used multiple versions of Unet and a modified version of U2net Show less

  • Senior Machine Learning Engineer 3.0

    Jul 2023 - Jun 2024

  • Machine Learning Engineer

    Aug 2021 - Jul 2023
• 

  Google

  Jun 2024 - now

  Software Engineer (GOOGLE XWF VIA ADECCO)