Oct 2022 – Jan 2023 · 3 mos · Remote

Apr 2022 – Aug 2023 · 1 yr 4 mos · Bangalore · Hybrid

• Engineered a fine-tuned OpenAI model that dynamically generates product descriptions by analyzing images and extracting relevant attributes; boosted product listing efficiency by 40%.
• Deployed Google Ads API to analyze keyword search volumes and TextStat for readability scores, optimizing content strategy and boosting user engagement by 25% over a six-month period.

Oct 2021 – Nov 2021 · 1 mo

• Tags: Recommender Systems, SVD, Collaborative Filtering, Cosine Distance, Cold start problem, Recall
• Description:
• a.) Developed a book recommendation system for amazon customers using collaborative and content-based filtering by utilizing the ratings of books and user features.
• b.) Implemented collaborative-based filtering using the KNN algorithm and created the sparse matrix to obtain user-item interactions and employed cosine distance to find user-item similarities.
• c.) Built an SVD model-based filtering and performed cross-validation and achieved recall@5 of 37.8% and recall@10 of 43.4%.
• d.) Implemented the solution for the cold start problem based on demographic-specific and weighted-rating book popularity and improved the efficiency of the user recommendation engine by 33%.

Sep 2021 – Oct 2021 · 1 mo

• Tags: SMOTE, XGBoost, Information Gain, Feature Engineering, Recall
• Description:
• a.) Developed a binary classification model using algorithms such as XGBoost, Stacked Classifiers to capture the underlying patterns and finally predict whether a company will go bankrupt.
• b.) Understood the correlation of features by using heatmap after capping outliers. Performed EDA and extracted important insights using various visualization techniques.
• c.) Handled data imbalance using various techniques such as Tomek Links and SMOTE-ENN and normalized the necessary features to ensure smoother model performance.
• d.) Experimented while creating various models combining with feature selection and sampling and achieved precision of 88.3% and recall of 86.6%.

Aug 2021 – Sep 2021 · 1 mo

• Tags: Regression, XGBoost, PCA, R2-Score, Multicollinearity, Outlier Handling
• Description:
• a.) Developed a regression model using algorithms such as Ridge Regression and
• XGBoost to predict the hourly demand for bikes and reduce the public waiting time
• b.) Designed the data pipeline to work on, understanding the impact of features such as time of the day, weather condition, and engineered important features for the model.
• c.) Imputed missing values, encoded categorical columns, handled outliers, checked multicollinearity, and performed PCA to reduce the dimensionality.
• d.) Experimented with hyperparameter tuning techniques such as Random Search and Bayesian Optimization and achieved an R2 score of 87% using the XGBoost model and reduced the public waiting time by 10%.

Jun 2021 – Apr 2022 · 10 mos

Jun 2021 – Jul 2021 · 1 mo

Jan 2020 – Mar 2022 · 2 yrs 2 mos · Pune, Maharashtra, India

• 1. Analyzing existing data sources for quality check and treating missing values.
• 2. Complied data for regularly scheduled reports & analysis
• 3. Support data collection, validation, cleansing, and analytics activities and provide recommendations for data quality improvements
• 4. Created UI/UX interface with internal and external customers.