Yogendra Yatnalkar đź’ˇ #

Connect/Follow Over:

About Me: #

I am a Senior Machine Learning Engineer with over three years of industry experience, working mainly on AWS cloud and partly on GCP cloud. In my current organization, I play three roles:

  • ML Engineer: I develop custom ML/DL solutions for global clients.
    • Have hands-on experience with:
      • Tabular-Data: Churn Prediction, Classification, etc
      • Breadth and Depth of Computer Vision domain: Segmentation, Object Detection, Image Similarity, Video Deduplication, Video Analytics, etc.
      • MLOps: SageMaker Jobs, SageMaker Pipelines, Airflow, Building Components like: Processing, Training, Monitoring, Explainability, etc
      • NLP: Currently building LLM and RAG based chatbots using AWS Bedrock and AWS OpenSearch.
  • Research Engineer: I primarily focus on Computer Vision research and have two main activities:
    • I explore new techniques and papers and try to see if they can be applied to any of the current projects (i.e., I try to find applications of the latest papers).
    • I explore new AWS ML services and evaluate their performance and usability.
  • Pre-sales ML Consultant: I also assist the sales team in my organization by interacting with customers, understanding their problems, and proposing possible solutions, especially for Computer Vision use cases.

Few Technical Achievements: #

  • I developed and deployed a custom GAN model for “Shadow Generation below objects” in early 2022.
  • I was one of the core developers of Quantiphi’s MLOps product on AWS : “NeuralOps”.
  • Conducting research on:
    • Backtracking AWS Lookout for Vision (I was recognized by AWS Community Builders for this)
    • Multi-Task Learning
    • Prompt-less Finetuning of MetaAI Segment Anything model (SAM)
    • Balancing Loss impact on ML training where multiple loss metrics are used simultaneously
    • ….. and many more
  • Experience on GPU:
    • Training a salient image segmentation model for seven days on a single GPU in 2021
    • Exploring Multi-GPU training with Horovod, Pytorch DP, Pytorch DDP, etc
    • Hosting large ML models on AWS Inferentia
    • Using FFMPEG on GPU for large-scale audio-video transcoding.
    • …. and many more
  • Working at scale:
    •  I worked on 1.5 TB of CSV data for churn prediction use case.
    • Worked on 70+ TB of raw audio-video data for content-deduplication use case

More About Me: #

Hey there, if you have reached till the end then you can visit the following pages: