Prince Ghosh Chowdhury

Key Skills:

Google Cloud Platform

Amazon Web Services

Airflow

Machine Learning

Deep Learning
AutoML & Custom Training
Terraform & Github Actions CI/CD
GPT2 & Whisper Large V2 Models
Batch Predictions, Online Prediction & Cloud Functions
Performance Evaluation & JSON Data Normalization
Google Kubernetes Engine

Professional Experience:
❖ Developed and deployed ML models using AutoML and custom training.
❖ Automated deployment processes with Terraform and Github Actions.
❖ Deployed GPT2 and Whisper Large V2 models for online and batch prediction - Improved text generation performance for real-time and batch predictions.
❖ Designed cloud functions for triggering batch inference tasks.
❖ Conducted performance evaluation and JSON normalization - Streamlined caption generation and improved media asset processing.
❖ Prepared data for training Google's Chirp model.

❖ Created dashboard with metrics for CaptionAI project.

❖ Experimented with different machine types for model inferencing on Google Cloud Platform - Optimized resources , saving time and costs for future inference tasks.

❖ Developed a JSON Normalizer for multiple STT models - Implemented the normalizer using OOP and python to reduce manual post-processing efforts.

❖ Used Terraform to automate the creation of GCS bucket.

❖ Deployed Whisper Large V2 model for batch and online predictions

❖ Deployed Pyannote model and integrated it with CaptionAI - Improved speaker diarization in transcription tasks.

❖ Trained and deployed YOLOv8 based sematic segmentation model to development of object detection features.

❖ Performed load testing for Speech-to-Text Endpoint.

❖ Fine-tuned Whisper model on legacy Discovery content using H100 nodes.

❖ Integrated Whisper_TS into the STT pipeline for batch predictions.

❖ AdReady - Owned a project called AdReady where the primary focus was to build the entire Machine Learning pipeline to process video assets and generate admarkers and push them to CAMP.

❖ Apache AIrflow - Used AIrflow to create a DAG to monitor the states of all the DAGs(queued, failed) and created a custom metric on Google Cloud Monitoring to visualize this data.

Certifications:
Google Kubernetes Engine Specialization - [Coursera]

Go Language

At WBSEDCL, I specialized in power distribution systems, gaining expertise in grid management, network planning, and asset management. I ensured regulatory compliance, conducted maintenance, and troubleshooting. Collaborated with teams on project execution, enhancing my skills in electrical engineering and project management within the energy sector.