I’ve worked in data-driven environments that required strong technical skills, collaboration, and accountability. These experiences have strengthened my ability to analyze data, communicate insights, and deliver meaningful results.
Data Analyst | University of Illinois at Chicago
Mar 2024 - Present
- Analyzed a dataset of 5,900+ refugee records with 435 variables from the OPENICPSR database to study employment outcomes among disabled refugees.
- Performed data cleaning, transformation, and recategorization to prepare a refined sample of 1,273 records for analysis.
- Used Power BI to visualize descriptive statistics and demographic distributions across gender, education, age, and disability status.
- Conducted exploratory data analysis (EDA) to examine factors such as education, English fluency, healthcare access, and public assistance.
- Built logistic regression models to predict employment likelihood using demographic, human capital, and social support variables.
- Evaluated model accuracy with precision, recall, and F1-score, achieving over 75% predictive reliability.
- Identified significant predictors including gender, age, and health insurance type that influenced employment probability.
- Derived data-driven insights to support inclusive policy recommendations for improving employment opportunities among disabled refugees.
Data Analyst Intern | CCC Intelligent Solutions
Dec 2024 - May 2025
- Retrieved and analyzed 5,000+ open-source car damage images, identifying severity patterns across 6 damage types and vehicle categories using Python and Excel.
- Created interactive Tableau dashboards to visualize image trends, supporting targeted synthetic data generation for computer vision modeling.
- Cleaned and curated dataset by filtering 1,200+ low-quality images and selecting 300+ representative samples per damage type to optimize training data quality.
- Generated structured captions using Ollama 3.2 vision model, capturing vehicle details to support downstream
fine-tuning and multi-model prompt development.
- Structured 1,800+ image–caption pairs for model ingestion using automated scripts, enabling efficient prompt-based fine-tuning and evaluation workflows.
- Fine-tuned Hugging face Stable Diffusion 1.5 via LoRA in PyTorch to train expert models on individual damage types, building
automated pipelines for scalable data ingestion.
- Collaborated with cross-functional teams to integrate expert models and prototype merged generation techniques for complex multi-damage scenarios.
- Evaluated models using Fréchet Inception Distance (FID), tracking improvements across training iterations and
delivering results via client-facing dashboards.
Business Analysis using Spreadsheets Instructor | University of Illinois at Chicago
Aug 2024 - Dec 2025
- Conducted weekly workshops for 30+ students on Power BI and Tableau, focusing on real-world dashboards for KPI
tracking and employee performance.
- Designed hands-on SQL and Excel assignments on workforce analytics and attrition trends, reinforcing applied data modeling skills for business insights.
- Evaluated student assignments and delivered actionable feedback, enhancing their grasp of data-driven decisionmaking and analytical storytelling.
- Offered personalized support to students on SQL, dashboarding, and insight generation, contributing to improved assignment performance and engagement.
- Managed course materials and datasets via Confluence, ensuring structured and consistent student access
throughout the semester.
- Resolved technical issues related to Power BI, Tableau, and SQL Server, ensuring uninterrupted learning and
hands-on tool usage in class sessions.
- Partnered with faculty to iteratively update teaching content based on student progress, integrating real-world use cases to enhance relevance.
<aside>
[**Home**](<https://necessary-quark-0ca.notion.site/Sai-Deepak-s-Portfolio-2e8174c2c52d8067aec8c582e216b1bb>)
</aside>
<aside>
[**Education**](<https://necessary-quark-0ca.notion.site/Education-29f174c2c52d8082838ae1143ce9c9e4>)
</aside>
<aside>
**[Skills & Certifications](<https://necessary-quark-0ca.notion.site/Skills-2e6174c2c52d8057abbcc2e4ec239bc8>)**
</aside>
