Open to opportunities

Anurag Sogani

@anuragsogani

Message

Experienced Data Scientist with expertise in keyword planning and credit risk modeling.

India

Message

What I'm looking for

I am looking for a job that offers a challenging and dynamic environment where I can apply my expertise in data science and machine learning. I value a company culture that encourages innovation and collaboration. I am particularly interested in roles that involve keyword planning, credit risk modelling, and predictive analysis. I am also open to exploring new domains and expanding my skill set.

I am an experienced Data Scientist with a strong background in keyword planning and credit risk modeling. In my current role at Datawrkz, I have engineered a proactive keyword planning solution that has achieved an 85% average forecasting accuracy. By employing various forecasting models, including the Random Forest algorithm and XGBoost, I analyze over 10,000 data points weekly to predict keyword performance. Additionally, I have incorporated search engine trends and social media analytics as additional data sources, resulting in a 15% enhancement over baseline models.

During my time at MyShubhLife, I enhanced the accuracy of the CRM predictive model by 10% using the Random Forest algorithm on NTC users' bank statement data. I also achieved a 50% increase in click rate through precise targeting using the Random Forest algorithm on promotional-marketing SMS data. I have a strong track record of collaborating across teams to validate model performance, optimize hyperparameters, and provide insights to refine credit risk assessment.

Prior to my current roles, I worked as an ML Engineer at Bitwise Academy, where I crafted a smart Discord chatbot in Python and Rasa. I also collaborated with the content team to curate training data for Rasa, enhancing the chatbot's responsiveness to user queries effectively. As an intern at Axis India ML Research Lab, I developed a generalized pre-processing framework for streamlined data transformation and implemented image augmentation techniques to optimize CNN model performance.

Experience

Work history, roles, and key accomplishments

Current

Data Scientist

Current

Datawrkz

Oct 2023 - Present (2 years 8 months)

Engineered a proactive keyword planning solution, achieving an 85% average forecasting accuracy validated against historical data. Employed various forecasting models, including the Random Forest algorithm and XGBoost, to analyze over 10,000 data points weekly, predicting keyword performance. Incorporated search engine trends and social media analytics as additional data sources, resulting in a 15

Python SQL XGBoost Word2Vec AWS S3 Athena QuickSight scikit learn TensorFlow ETL CI CD Data Analysis Marketing AWS EC2 AWS Glue AWS Sagemaker Data Science Machine Learning

Data Scientist

MyShubhLife

Apr 2022 - Oct 2023 (1 year 6 months)

Enhanced CRM predictive model accuracy by 10% using the Random Forest algorithm on NTC users' bank statement data through meticulous feature engineering. Achieved a 50% increase in click rate through precise targeting using the Random Forest algorithm on promotional-marketing SMS data.

ML Engineer

Bitwise Academy

Aug 2021 - Mar 2022 (7 months)

Crafted a smart Discord chatbot in Python and Rasa, integrating natural language understanding to improve handling of course-related questions.

Python scikit learn

Intern

Axis India ML Research Lab

May 2019 - Dec 2019 (7 months)

Developed a generalized pre-processing framework for streamlined data transformation and reusability. Implemented image augmentation techniques to optimize CNN model performance. Designed data pipelines leveraging NumPy, pandas, scikit-learn, and TensorFlow for efficient preprocessing and scaling operations.

Python NumPy Pandas scikit learn TensorFlow

Current

Advanced Data Science Associate

Current

ZS Associates

Jun 2024 - Present (2 years)

Built a production RAG chatbot over 500+ unstructured documents (PPTs, call transcripts) using Azure OpenAI GPT-4, achieving 90%+ query accuracy with sub-2s semantic retrieval. Implemented the end-to-end RAG pipeline in Dataiku DSS and deployed on Azure, adopted by 3 brand teams and reducing manual research time by 60%; also engineered an NBA model that delivered +256% more new prescribers.

RAG Dataiku Python Snowflake Databricks LightGBM SHAP Explainability MLOps