Hello, my name is Zaeem. I'm a Computer Science major at the University of Massachusetts Amherst, with a focus on artificial intelligence and security.
This winter break, I interned at Microsoft on the Health and Life Sciences team, building an AI assistant that helps radiologists speed up and simplify parts of their reporting workflow. Last summer, I interned at Marriott International, where I built a chatbot that lets people ask data questions in plain English and get clean, SQL-backed answers.
On campus, I have done research at CIIR on making LLMs more personalized and more useful over time.
When I'm not coding, I'm probably watching football, cricket, tennis, or Formula 1 and pretending it's data analytics!!
Experience
Software Engineer Intern
Microsoft
- Worked with the Health and Life Sciences team to build a cloud-based assistant that automates diagnostic radiology workflows.
- Built an AI radiology assistant using FastAPI, GPT-5.1, and spaCy, automating finding extraction and reducing report time by 80%.
- Engineered DuckDB vector search mapping findings to 2,150+ codes, eliminating manual lookup for medical documentation.
Software Engineer Intern
Marriott International
- Built an internal chatbot in Microsoft Power Apps that converts natural language to SQL via GPT-4, built on a Power BI semantic layer.
- Executed auto-generated SQL across 7 domains in Snowflake with RBAC and audit logs, returning results with sub-15-second latency.
- Delivered actionable, SQL-transparent data insights with 86% accuracy, logging lineage and metadata in Azure for auditability.
Software Development Extern
Snap Inc.
- Designed a soccer-themed AR lens incorporating Snap's Lens Studio and 3D modeling, certified by Snap Inc.'s Head of Entertainment.
- Launched a Snapchat lens inspired by Reebok and soccer, compatible with iOS and Android, gathering views from 100+ countries.
- Conducted data-driven market research in sports and technology using Tableau and MySQL, enhancing data visualization by 30%.
Data Science Intern
Neftwerk
- Designed and sorted data from CSVs to Attio dashboards to improve client information accuracy and operational efficiency by 60%.
- Leveraged Attio's REST API to implement filtering and pagination, optimizing data retrieval by 30% and enhancing user experience.
- Engineered Python automation for Excel-to-CSV pipeline, leveraging Git version control to optimize Attio dashboard data processing.
AI and LLM Research Intern
Center for Intelligent Information Retrieval
- Built an LLM personalization platform using AWS and vLLM; rotated OpenRouter models like Claude 3.5 to analyze user preferences.
- Captured 100 preferences per user in DynamoDB and streamed them to a DPO pipeline, improving the personalization benchmark.
ML Research Intern
University of Massachusetts Amherst
- Optimized a CNN for ASL recognition, achieving a 97% F1 score through data augmentation and additional convolutional layers.
- Utilized ResNet pre-trained models to enhance accuracy to 98% and reduce overfitting, improving generalization on unseen data.
- Enhanced model robustness in Keras with batch normalization and the Image Data Generator for consistent validation performance.
Projects
Each project has a link to my GitHub code. Feel free to look at the quick video demos or check out my code.
PerfectPitch
AI-powered platform designed to provide tailored feedback, script assistance, and a performance score for personalized interview preparation.
LeetBank
Web-based platform designed to help you store, annotate and manage your LeetCode questions efficiently.
Uber Data Pipeline
A scalable pipeline using Python, GCP, Mage.ai, and BigQuery to process NYC Uber trip data, automating ETL and delivering insights via Looker Studio.
PomoPay
A Pomodoro-based productivity app that sets weekly goals, links payments to accountability, and boosts focus with secure Stripe integration.
Sonar Data Classification
Developed a sonar signal classification model to distinguish between "Rock" and "Mine" signals using XGBoost.