Data Analyst Computer Vision

Vimoli Mehta

Austin, TX +1-737******* ***********@******.*** Linkedin

SUMMARY

A data enthusiast with strong analytical skills, 2+ years of knowledge in SQL, Python, and R, and a knack for translating technical insights into actionable business solutions through data storytelling. EXPERIENCE

Data Analyst Intern, Clean Harbors, Norwell, MA May 2023 – Aug 2023

• Enhanced operating efficiency by 20% by defining critical KPIs for measuring trapped inventory at distribution centers (DCs) and developing an interactive Power BI dashboard, consolidating real-time data from multiple sources.

• Forecasted future inventory needs and prevented overstocking through the design of predictive models.

• Spearheaded the development of a Python script to monitor daily stock discrepancies across multiple business units, processing data from 80k records, resulting in time savings of over 10+ hours per week for the team.

• Improved supply chain visibility by executing complex SQL queries for insightful ad hoc reports in SQL Server Management Studio.

• Employed statistical modeling techniques to assess the cost-effectiveness of in-house electronic recycling, in close collaboration with cross-functional teams. The prototype holds the potential to yield a 12% reduction in recycling cost. Teaching Assistant, University of Texas at Austin, Austin, TX Jan 2023 – Present

• Delivered personalized guidance to 50+ students for course I306A: Statistics for Informatics and collaborated with the professor to generate course material and R programming modules along with grading. Data Science Intern, Softvan Pvt.Ltd, Ahmedabad, Gujarat Jan 2022 – May 2022

• Developed, fine-tuned, and deployed a computer vision model for employee activity detection, achieving a 92% accuracy.

• Annotated and cleaned dataset of 10,000+ images using CVAT, reducing false positives by 15%, and increasing recall by 10% through hyperparameter refinement.

• Integrated the computer vision model into a custom Flask-based web application using TensorFlow and YOLO object detection, enabling real-time monitoring for 50+ employees, and enhancing data-driven decision-making. Data Visualization Intern, Saint Louis University, GlobalShala, Jan 2022 – May 2022

• Conducted in-depth analysis of the "Superhero U" event's Facebook ad campaigns, evaluating key performance metrics including CTR(click-through rate), conversion rates, and CPC(cost-per-click).

• Created interactive Tableau dashboards, facilitating quick insights and improving decision-making for stakeholders.

• Recommended discontinuing 3 underperforming ad campaigns based on rigorous statistical analysis, resulting in a 5% reduction in company costs while maintaining or enhancing conversion rates. TECHNICAL SKILLS

Languages: Python, SQL(Postgres, MySQL), R, C, C++, Bash(Unix Shell), Git, Javascript, Django, Flask Libraries: NumPy, Pandas, Seaborn, Matplotlib, SciKit-learn, SciPy, NLTK, Spacy, OpenCV, PyTorch, ggplot Tools: Excel(Pivot, VLOOKUP, VBA), Alteryx, Power BI, Tableau, Weka, Microsoft Access, Outlook, JIRA, Snowflake Cloud: Microsoft Azure, Google Cloud Platform (GCP), Amazon Web Services (AWS) EDUCATION

The University of Texas at Austin, Austin, TX Aug 2022– May 2024 Master of Science in Information Sciences– Data Science Coursework: Data Storytelling, Predictive Analysis & Data Mining, Machine Learning, Data Science Lab, Python, databases & Big Data Institute of Technology, Nirma University, Gujarat July 2018 – May 2022 B. Tech, Electronics, and Communication Engineering, Minors in Computer Science Coursework: Data Structures, Linear Algebra, Econometrics, Vector Calculus & Probability Distribution, Operating Systems PROJECTS

Kaggle Competition, Course Project [Link] 2022

• Performed extensive data preprocessing, including outlier handling, missing value treatment, one-hot encoding, and advanced feature scaling and dimensionality reduction to enhance data quality.

• Implemented 6+ ML models and performed hyperparameter tuning (GridSearchCV and Randomized Search), alongside advanced ensemble techniques (stacking, ensembling, and blending), with AUC as the evaluation metric – CATBoost performed best. Song Lyrics Generation 2022

• Gathered data from 20k songs through web scraping and cleaning to later use songs from 11 core genres. Evaluated 3 NLP models

(GPT-2, BERT, LSTM) for song lyric generation, with GPT-2 delivering a 20% improvement in BLEU metric over other models. Employee Attrition System Using Tree-based Ensemble Techniques [Link] Published in C2i4-2021 on IEEE Explore 2022

• Developed an advanced employee churn prediction model by trying various supervised ML models along with stacking, ensembling, and feature engineering, achieved an accuracy (95.05%)- higher accuracy than the currently existing models using GradientBoost classifier and 85.37% through Random forest.

Contact this candidate