Vijay Kumar

Client: Woodside Energy

• Trained Yolov8 model on custom labels i.e fire and smoke to help identify and prevent any disaster in lng plants.
• Worked on identifying machine tags from pdf docs and image files using python and deploying the solution in aws using lambda.
• Researched on image depth map use case. Used different models of MiDas and analysed the results.
• Worked on predicting optimum lng gas rate using classical machine learning algorithms.

Client: JPMorgan Chase

• Built keyword extraction and theme recommendation solution using average word2vec, cosine similarity and k-means clustering.
• Build custom logic for Emerging theme detection to be used as part of MLOps, post model deployment to production.

• Built the Anomaly detection solution using Amazon lookout
• Created the entire architecture for the solution
• Used s3 bucket as a data source
• Created Aws lambda functions using SAM and ECR for sending custom alerts on Slack
• Product: Customer Call Analysis Working on Speaker Identification using GCP api, Sentiment analysis, Emotion detection, Hold time detection , Competitor Analysis, Audio pitch etc for customer call audio data
• Using Flask for api development
• Created docker image for deploying on GCP instance