About
I'm an incoming Software Engineer at Microsoft and a graduating Computer Science student at Northeastern University. I enjoy full-stack engineering work that turns complex problems into products people can use every day.
Most recently, I interned on Microsoft's Copilot for Sales team, where I shipped an AI-powered meeting insights feature in Outlook. The feature helped sales teams reduce pre-meeting research time and prepare faster.
Outside of work, you can usually find me watching movies, playing tennis, skiing, or planning my next trip. Always looking for a good recommendation.
Experience
Software Engineer Intern ↗
Led the design and launch of a Copilot for Sales meeting insights feature that generated AI-powered briefs directly in Outlook. Built responsive front-end experiences with React, Redux, and Fluent UI, and developed scalable C# APIs with ASP.NET Core. Deployed to Azure Kubernetes Service through Azure DevOps CI/CD pipelines.
Software Engineer Co-op ↗
Developed Python and FastAPI serverless APIs for AI recommendation systems handling 1,000+ requests per minute with sub-100ms latency. Built AWS ETL pipelines processing 100GB+ daily and improved processing speed by 40% through caching and parallelization.
Security Engineer Co-op ↗
Built Python automation integrating Splunk and CrowdStrike to streamline incident triage, reducing response time from 4 hours to 15 minutes. Designed graph-based workflows to cross-reference 500M+ compromised credentials, strengthening proactive threat detection.
Projects
NYTimes Pips AI Solver ↗
Built an AI agent to solve the NYTimes Pips puzzle using two complementary approaches: a constraint satisfaction solver with backtracking and a simulated annealing local search solver.
Kambaz Learning Management System ↗
Developed a full-stack learning management system using Next.js and MongoDB, supporting 50+ courses, assignment workflows, and three user roles.
NLP Emotion Analysis ↗
Trained and benchmarked four emotion-classification models on 20,000+ labeled tweets: Naive Bayes, Logistic Regression, Bi-LSTM, and fine-tuned DistilBERT. DistilBERT achieved a 92% F1 score across six emotion categories, outperforming baselines by 15%+.
visuaLAG (HackMIT 2023) ↗
Built a command-line tool at HackMIT 2023 that visualizes and gamifies LeetCode execution. Designed a staged learning flow so users can build algorithm intuition through interactive, visual feedback instead of only pass/fail test results. Built it to be language-agnostic so it can visualize execution independent of implementation language.
BattleSalvo ↗
Developed a Java battleship strategy game with both single-player (AI) and multiplayer modes. Implemented an AI opponent for ship placement and shot selection, and built a real-time client-server socket layer to support multiplayer gameplay.
GiveBack (TreeHacks 2023) ↗
Built a donation platform at Stanford's TreeHacks 2023 using React and Python, connecting users with 100+ nonprofit organizations. Developed donation-tracking features, including leaderboards and personalized user views, and implemented an NLP-based matching algorithm to align user preferences with relevant nonprofits.