hi! i'm nathan 👋
i'm an undergraduate at uc berkeley (go bears!) studying computer science & data science. my academic interests lie in machine learning, full-stack web development, and data engineering pipelines.
currently, i'm focused on building scalable cloud-native architectures, designing data-driven systems, and looking for internship opportunities. beyond writing code, i'm an avid photographer capturing urban structures and natural landscapes.
in my free time, i like to go on a sidequest with my camera, hit the gym to clear my head, or talk about JDM cars.
Data Engineer Intern
May 2025 - Aug. 2025- Built a real-time LLM decisioning pipeline on AWS (Kinesis, Lambda, Step Functions) to process 100k events/day which enforced schema-validated JSON outputs and wrote full traces to S3 for auditability and replay.
- Improved platform reliability by adding idempotency keys in DynamoDB and CloudWatch alarms to reduce failed executions and eliminated release-related deployment failures by 33%.
- Cut dashboard data retrieval time by 80% by rewriting high-cost joins, adding composite indexes, and implementing cursor pagination in Django/PostgreSQL APIs.
Software Engineer
Jan 2025 - Aug. 2025- Prototyped a mobile CI/CD pipeline with GitHub Actions, automating deployments to TestFlight + Google Play, which cut manual release times from hours to minutes.
- Deployed a native-feel, in-app chat interface that integrated with Google’s Dialogflow API to provide real-time AI support, reducing manual support ticket volume by 33%.
- Built a Django backend and admin CMS that enabled marketing to update in-app content without requiring an app store submission.
Data Science Intern
Jun. 2024 - Aug. 2024- Built a near-real-time SAP inventory sync for 3,647+ SKUs using incremental pulls and upserts into Postgres which reduced manual entry by 80% and enabled automated invoicing for 100% of customers.
- Rebuilt a legacy platform (Django REST, JWT, React) and improved performance 80% by reducing payload size and adding server-side caching.
- Built a Docker-based CI/CD workflow to deploy ETL/ELT jobs automatically, streamlining legacy database migration and improving pipeline reliability.
Flow
ActiveUnified full-stack email and calendar mobile app utilizing AI sorting, multi-account syncing, and metadata indexing.
Clickbait Classifier
CompletedFine-tuned a BERT-based classifier on 18K social-news titles; achieved 0.89 F1 and 91% accuracy on a held-out test set, outperforming a TF-IDF logistic regression baseline by +0.14 F1 points.
ItineraryAI
CompletedBuilt a chat-based travel concierge that generates structured itineraries from a single prompt and shipped a web dashboard for trip planning (budgets, maps, calendar sync).
The IBD Digest
CompletedBuilt an offline-first IBD companion app for 5,000+ users (recipes, ingredient libraries, SIBDQ assessment) with persistent onboarding and local state persistence.
SimplyMail
CompletedModern web Gmail client focused on high performance, clean visual interfaces, and secure Gmail API integrations.
Spotify Analytics
CompletedPersonal listening telemetry dashboard showing profile insights, favorite genres, and user recommendations.
University of California, Berkeley
Class of 2026GPA: 3.84
Focus: Machine Learning, Software Engineering, & Database Architectures.
Computer Science
Data Science
Engineering & Analytics
Physics
Why I Love Data Science
April 2, 2025A reflection on why data science excites me, statistical insights, and the power of data visualization.
Read PostMy Gym Routine
May 20, 2025My weekly gym routine, tracking consistency, personal milestones, and how working out helps clear my head from coding.
Read PostHow I Built My Portfolio Website
March 24, 2025A breakdown of my serverless tech stack, custom Durable Objects trackers, R2 integration, and design decisions.
Read Post