I’m Lina, a computer science student at Northeastern interested in software engineering, machine learning, and interpretable AI.
I build full-stack applications and explore how intelligent systems make decisions. I’m especially interested in making model behavior understandable and inspectable, from recommendation algorithms to reinforcement learning agents.
My experience includes international logistics software, neural network development, and interactive geospatial research. Outside computer science, I make pottery, play guitar, climb, and bake banana bread. I love spending time with friends and getting deep into books, music, movies, and art.
A reinforcement learning simulation for exploring budget-constrained transit improvements.
How can changes to a transit network reduce average commuter travel time without exceeding a budget?
Designed a custom Gymnasium environment with graph-based state representations and trained a DQN agent to modify network edges.
Used replay buffers, target networks, epsilon-greedy exploration, and reward shaping to support learning. Visual evaluations make network changes easier to inspect.
A simulation for evaluating transit decisions and examining an agent’s behavior under network and budget constraints.
An inspectable recommendation simulator showing how interactions shape a feed.
Users cannot easily tell which interactions drive their recommendations or whether their feed is becoming less diverse.
Developed a full-stack simulator with feed audits and recommendation traces. REST endpoints compute diversity and relevance metrics, with pytest coverage for audit calculations and error handling.
Kept metrics and ranking explanations in deterministic code. Used Claude to interpret audit results in natural language, while traces expose the interaction history and interest profile behind a ranking.
Users can inspect both patterns in their feed and the simulator’s actual ranking logic, with a clear separation between computed evidence and AI-generated interpretation.
A modular C++ effects pipeline for processing live guitar audio.
Interactive audio effects need responsive controls while continuously processing a live signal.
Integrated a modular pipeline for guitar input and stereo output, with delay, distortion, and reverb, adjustable parameters, presets, and WAV export.
Used thread-safe parameter updates so controls can change during audio processing. A terminal interface displays the active settings.
A playable real-time effects workflow with configurable processing and saved audio output.
From logistics software to historical data, I’ve worked on systems that connect information to people.
Created five interactive geospatial dashboards exploring historical policy impacts in Cambridge. Developed a virtual exhibit with the mayor’s office, bringing archival materials and ArcGIS maps into one interface.
Improved SaaS applications for international logistics across the frontend and backend. Worked on routing algorithms and API integrations connecting USPS sorting hubs and freight forwarders using C#, .NET, SQL Server, and JavaScript.
One of four local Student Leaders for an eight-week internship. Led tech workshops for 30 students, presented a strategic proposal to the museum’s executive board, and joined a national leadership summit.
Taught Python and Java, organized a free summer camp and GirlsHack, and led a five-person web team publishing community stories and event photos.
B.S. Computer Science · AI concentration
September 2023 – December 2026 · Dean’s List
⠀⠀⠀⠀⠀⠀⠀⠀⢀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⣰⠋⣷⠀⠀⠀⠀⠀ ⠀⠀⢀⡼⢲⠀⠀⠀⠀⠀⠀⠀⠀ ⢀⡖⢳⠄⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⣸⠃⢀⣿⠖⢳⡆⠀⠀⠀⢀⡞⠀⣸⣠⠤⡄⠀⠀⠀⠀⢠⡟⠁⣸⣤⠶⡄⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⢰⡇⣴⡟⠁⠀⡼⠁⠀⠀⠀⡾⢀⣴⠏⠀⢠⠇⠀⠀⠀⠀⡾⢀⣴⠋⠀⢰⠇⠀⠀⠀⠀⠀⠀ ⠀⠀⣀⠴⠋⠉⠘⢹⡇⢀⡼⠳⠤⣤⡤⠒⠣⢟⡇⠀⢠⣏⣀⡀⣠⠴⠚⠣⢯⡇⠀⣠⢯⣄⣀⠀⠀⠀⠀⠀ ⠀⡼⠃⢀⣄⠀⠀⠘⠓⠋⠀⢠⡞⠁⠀⠀⠀⠘⢧⠴⠋⠀⢠⠟⠁⠀⠀⠀⠘⠧⠞⠁⠀⠀⠈⠙⢦⡀⠀⠀ ⢸⡇⠀⠈⠉⠀⠀⠀⠀⠀⢠⣏⡀⠐⠟⠃⠀⠀⠀⠀⠀⢠⣏⠀⠘⠛⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠹⡄⠀ ⠀⠙⠦⣤⣀⣤⠤⠀⠀⠀⠈⠳⣄⡀⠀⠀⢀⠀⠀⠀⠀⠀⠻⢆⣀⢀⡀⣀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⣿⠀ ⠀⠀⠀⢻⡀⠀⡀⠀⠀⠀⠀⠀⠀⢻⠉⠉⠁⠀⠀⠀⠀⠀⠀⠀⢹⡉⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⡿⣄ ⠀⠀⠀⢨⠇⢀⡿⠀⠀⢀⣀⡤⠄⠈⣧⠀⢲⠀⠀⠀⠀⢀⡀⠀⠈⣷⠀⢸⡀⠀⠀⠀⣀⡀⠀⠀⠀ ⠀⣡⠟ ⠀⠀⠀⠙⠒⢯⣀⣠⠴⠻⢧⣄⠤⠼⢥⣴⠋⠀⢀⡴⡟⠉⢀⣀⣸⡧⣴⠋⢀⣀⡴⣟⣁⣀⣀⡠⠴⠊⠁⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠙⠉⠁⠀⠉⠉⠉⠀⠀⠀⠈⠉⠉⠀⠀⠈⠉⠁⠀⠀⠀⠀⠀⠀
Open to talk!
I’m looking for AI and software opportunities starting January 2027.
