Badges
Certifications
tanishqarora2001 has not earned any certificates yet.
Work Experience
Data Scientist
Inito• January 2023 - Present
Responsibilities o Perform EDA to explore new opportunities for product improvement using classical and deep learning algorithms. o Develop new features in the feature stores, write DBT (in SQL) for new features created and push to Airflow ELT System. o Deploy new models into production along with all relevant test cases and all the preprocessing pipelines. Projects o Super Batch Detection and Classification Model (OpenCV| Yolo V8| Docker | GCP | PyTorch | GIT | CI/CD | Vertex AI) Engineered a model for super batch detection and classification (64 classes) using YOLO V8 & transfer learning. Established optimal confidence and Non-Max Suppression thresholds for production, resulting in a 99.98% accuracy rate. Containerized, deployed, and operationalized the model on Vertex AI for production, determining a requirement of atleast 8 machines. Developed a Flask server for local model testing, optimizing it for processing 30,000 images in 15 minutes using distributed computing. Developed CI/CD for automated deployment of model with 12+ test cases (Model Version Check, Model Availability Check etc). o Airflow – Airbyte – DBT’s Infrastructure Development (Airflow | Airbyte | Docker | Python | Multiprocessing | Multithreading) Set up and optimized tabular data processing for 1,000,000 daily rows with Airflow and Airbyte for seamless synchronization. Utilized Data Build Tool (DBT) to generate structured dataset using 10 SQL scripts to transform raw data to feature store. Containerized the data transformation infrastructure, hosted it on Google VM’s, and authored 4 DAGs for workflow scheduling. Designed and scheduled 3 user-level and 1 cycle day-level feature stores as DAGs, optimizing processing time through parallelization. o User-Churn Prediction Model (Python | SQL | Bigquery | TSNE | UMAP | PCA| K-means | Multivariate Analysis) Conducted EDA on 50+ features to pinpoint the primary factor leading to customer product discontinuation. Performed Multivariate analysis, rigorously testing over 25+ hypotheses to unveil trends within various month user cohorts. Applied clustering techniques to categorize users based on their first 30 days activity and predict their LTV (Life Time Value).
Business Analyst
OLX• February 2022 - January 2023
Responsibilities o Managed end – to – end I2P (Inspection to Procurement) Funnel Analytics. o Analysing Gaps and Highlighting Leakages in Multiple New Pilots to Improve I2P, reduced gaps by 30%. o Developing machine learning models to enhance I2P (Inspection to Procurement) at every available opportunity. o Defining and Producing dashboards for tracking model performance, A/B Testing etc for effective decision making by Stakeholders. Projects o Lead Scoring Model (SQL | Python | Tableau | Jenkins | Feature Engineering | Regression) Reduced the incoming leads to RA by 12% by identifying junk leads, Reduced leads not reached by RA by 20%. Used features like lead source, prior funnel interactions, talk time/connects ratio etc (30+ hand engineered features). Validated credibility to be 95% by model and 5% because of RA Utilization by using Time Motion Study & output metric of RA. o Dealer Prediction Model (SQL | R | Python | Tableau | Jenkins | Tree Models) Leveraged car attributes (make, model, year) and dealer history to predict which dealer is most likely interested in buying a specific car. Achieved accuracy rates within no of calls as 27% (1st call), 32% (2nd), 53% (3rd), 65% (4th), and 72% (5th). Helped DRM to successfully identify potential buyers in the first 5 dealer calls (out of 6000 dealers).
Education
DTU, Delhi (Delhi Technological University - formerly DCE)
PIE, B.Tech• June 2017 - July 2021
JEE Advance 2017 AIR 7118
Links
Skills
tanishqarora2001 has not updated skills details yet.