Surya Teja Email & Phone Number

Columbia, South Carolina, United States

• Responsible for setting up technical and functional requirements, data pipelines, data preparation, design, development, modelling, testing and deployment of advanced models in the AWS cloud.• Conducted exhaustive study on various available Transformers models and estimated costing, system load, model performance, cost to performance metrics to decide on the best model that would yield the best results in terms of Development, Optimization and Maintenance parameters.• Exported data from Snowflake to S3 and created on-demand tables using AWS Lambda Functions and AWS Glue using Python and PySpark.• Designed and implemented ETL pipelines on S3 files on data lake using AWS Glue.• Leveraged Pandas, NumPy and other preprocessing libraries for Data Cleaning and Feature Engineering to incorporate the same into the pipeline component of the NLP workflow.• Worked with 7TB of Unstructured Data for a classification problem and built NLP Pipelines which includes preprocessing, modeling and testing components.• Performed text preprocessing like Tokenization, Lemmatization, etc. and Feature Engineering/Extraction on multi label imbalanced data followed by applying various sampling techniques.• Built SpaCy Huggingface transformers models like BERT, ALBERT and RoBERT with de-identified transcript.• Built metrics like Confusion Matrix and Model Performance via dashboards and integrated into sisence for real time visibility.• Finetuned BERT models on de-identified client data to streamline development & deployment in cloud.• Saved $500,000 in costs to the client by developing the advanced models for setting up processing in place for CI/CD.• Achieved an overall Model Accuracy of 93% which was a 25%-point increase from old models resulting in Significant Cost Savings to the client.Environment: AWS Cloud, Python, SQL, ETL, jupyter notebook, Pyspark, Pandas, Snowflake, Sagemaker, Huggingface, SpaCy, Transfer learning, Git, JIRA, Agile, Windows and Linux.

New York, United States

• Estimated Kickstarter project consistency at 7% based on project description by Conducting a Similarity Analysis in python.• Applied Text Pre-Processing Techniques on the data and did a similarity analysis on two of the key evaluation features which shows the project goals consistency on Kickstarter.• Estimated that 36% of reddit content was regarding health issues in 2020 in python using NLP Techniques like LDA, TF-IDF etc.• Applied text pre-processing techniques on the data and did Topic Modelling Analysis which gave a clear picture of the nature of content posted on reddit.• Developed automated scripts to Scrape Kickstarter and Reddit Text data and build a corpus.• Saved the web scraped data to GCS, and built ETL process to automated the data loading process.• Employed multi core processing and leveraged GPU to reduce latency while scarping data and save to the cloud and built Data Pipelines in Airflow in GCP.• Experience in building and architecting Multiple Data Pipelines, end to end ETL process for Data Ingestion and Transformations in GCP and coordinate tasks among the team.• Integrated Git with GCP for version control and collaborative development effort.• Leveraged Pandas, NumPy and other preprocessing libraries for Data Cleaning and Feature Engineering to incorporate the same into the pipeline component of the NLP workflow.• Performed data Pre-Processing and Vectorization using NLTK and Gensim libraries in python.• Built the data warehouse with 3TB of Unstructured Data for a semantic analysis problem and built NLP Pipelines which includes preprocessing, modelling and testing components.• Utilized Agile process and JIRA issue management to track sprint cycles and prepared documentation for future quantitative analysis.Environment: Google Cloud, Python, SQL, ETL, jupyter notebook, Pandas, Snowflake, Gensim, NLTK, SpaCy, Git, JIRA, Agile, Windows and Linux.

Tampa, Florida, United States

• Responsible for building the data warehouse, data pipelines, data preparation, design, development, modelling, testing, deployment and integration with dashboarding tool in the Google Cloud.• Developed automated scripts to scrape SEC 10k and YouTube data and build the data warehouse for companies in IT domain.• Saved the data to GCS, created tables and saved to BigQuery and built ETL process to automated the data loading process.• Employed multi core processing and leveraged GPU to reduce latency while scarping data and save to the cloud and built Data Pipelines in Airflow in GCP.• Experience in building and architecting Multiple Data Pipelines, end to end ETL process for Data Ingestion and Transformations in GCP and coordinate tasks among the team.• Developed python code for different tasks, dependencies, SLA watcher and time sensor for each job for workflow management and automation using Airflow tool.• Build data pipelines in airflow in GCP for ETL related jobs using different airflow operators incorporating old and newer operators.• Created BigQuery authorized views for row level security or exposing the data to other teams.• Experience in building multiple Data pipelines, end to end ETL and ELT process for Data ingestion and transformation in GCP and coordinate task among the team.• Design and implement various layer of Data Lake, Design star schema in Big Query.• Using g-cloud function with Python to load data into BigQuery for on-arrival csv files in GCS bucket.• Process and load bound and unbound data from Google pub/sub topic to BigQuery using cloud Dataflow with Python.• Developed and Demonstrated the POC, to migrate on-prem workload to Google Cloud Platform using GCS, Big Query, Cloud SQL and Cloud DataProc.Environment: Google Cloud, BigQuery, Python, SQL, ETL, jupyter notebook, Pandas, Snowflake, SpaCy, Git, JIRA, Agile, Windows and Linux.

Tampa, Florida, United States

• Extracted data from multiple source systems S3, Redshift, RDS and Created Multiple Tables/Databases in Glue Catalog by creating Glue Crawlers.• Used AWS Data Pipeline for Data Extraction, Transformation and Loading from homogeneous or heterogeneous data sources and built various graphs for business decision-making using Python matplot library.• Responsible for gathering requirements, analysis, design, development, model building, testing and deploying model in AWS Cloud Environment.• Used AWS Glue Catalog with crawler to get the data from S3 and perform SQL query operations and JSON schema to Define Table and Column Mapping from S3 data to Redshift.• In Tableau development environment, Supported Customer Service designing ETL Jobs, Dashboards utilizing data from Redshift.• Wrote scripts for Schema Validation, Transformations and Logging and connected the same to a dashboarding tool.• Built Random Forest on key features from patient data which Improved clinic Operating Efficiency by 38%.• Deployed the model using AWS Sagemaker and built a simple front end user interface to check the daily status of clinic performance.• Reduced the Insurance Processing Time by 7 days and improved the efficiency by 200%.• Used GitHub for version control and team collaboration.• Utilized Agile process and JIRA issue management to track sprint cycles.Environment: AWS EC2, Python, MySQL, jupyter notebook, Sagemaker, Ensemble Methods, Git, JIRA, Agile, Windows and Linux.

Hyderabad, Telangana, India

• Translate functional and technical requirements into detail Specifications running on AWS using services EC2, ECS, RDS Aurora MySQL, SQS, SNS, KMS, Athena.• Involved in gathering business requirements, logical modeling, database design, data sourcing and data transformation, data loading, SQL, and performance tuning.• Structured multiple projects in Object Oriented Programming in python, developed dashboards in Tableau and executed cloud instances.• Created Stored Procedures, Views and User Defined Functions to support the front-end application.• Experience in developing Spark applications using Spark-SQL in Databricks for data extraction, transformation, and aggregation from multiple file formats.• Preprocessed Data using pandas dataframes for generating throughput capacity, product to zip mappings based on inventory breadth constraints, initial inventory, and other flow related data.• Developed and maintained PAN-EU Inbound Analytics and Reporting tools that facilitated freight planning and scheduling.• Predicted labor productivity with 82% accuracy which led to Freight Optimization across regions in EU.• Identified 50% more KPI’s from queried data using SQL after Implementing Data Cleaning techniques in Python.• Collaborated with supply chain for inception and Deployment of 3PL sites with 92% success.• Concatenated Multiple Data Tables to forecast Supply Chain Dynamics (~27M orders) across various time frames.• Parallelized the Data Flow Leveraging AWS Cloud tools like EC2, redshift, kinesis, Glue etc. to build data pipelines and added logging functionality.• Used GitHub for version control, team collaboration and utilized Agile process and JIRA issue management to track sprint cycles.Environment: Python, jupyter notebook, Tableau, Pandas, numpy, JSON, CICD script, Agile, SQL, Unix, AWS Services – S3, EC2, Redshift.

Hyderabad, Telangana, India

• Used various AWS Services including S3, EC2, AWS Glue, Athena, RedShift, EMR, MS, Kinesis etc.• Extracted data from multiple source systems S3, Redshift, RDS and Created Multiple Tables/Databases in Glue Catalog by creating Glue Crawlers.• Used AWS Data Pipeline for Data Extraction, Transformation and Loading from homogeneous or heterogeneous data sources and built various graphs for business decision-making using Python matplot library.• Saved Over $90,000 in Costs by utilizing Lean-Agile methods and by leveraging data achieved a 15.3% reduction in rollout times.• Exported data from Redshift to S3 using AWS Glue Jobs, Pyspark Context, boto3 and repartition procedures to convert to parquet format using create dynamic frames.• Visualized the data using Seaborn and Matplotlib to understand the relation between features and analyzed the statistical distribution of data.• Designed and Implemented ETL Pipelines on S3 parquet files on data lake using AWS Glue.• Managed Storage in AWS using Elastic Block Storage, S3, created Volumes and Configured Snapshots.• Experience Configuring AWS S3 and their lifecycle policies and to backup files and archive files in Amazon Glacier.• Experience in Creating and Maintaining the Databases in AWS using RDS.• Utilized Agile process and JIRA issue management to track sprint cycles.Environment: Python, jupyter notebook, Pandas, Amazon Web Services (AWS), Git, JIRA, SQL, Agile, Tableau, Windows and Linux.

Master Of Science - Ms, Business Analytics And Information Systems

Bachelor Of Technology - Btech, Mechanical Engineering