Meta — Software Engineering Intern, Compute Infrastructure
Present
Infrastructure for large-scale computation and experimentation.
Now
A first-author manuscript on multivector retrieval is under review at NeurIPS.
Adjacent threads: training and inference efficiency, systems for experimentation, and the geometry of representation in retrieval-augmented settings.
Selected experience
Aisera — AI / ML Engineer Intern
Summer 2025
Ontology-aware neural search for retrieval-augmented systems; clustering and embedding pipelines; FAISS-style index tuning for semantic search.
Research
SSAIL, University of Illinois — with Minjia Zhang
Ongoing
Multivector architectures and the systems infrastructure required to operate them reliably at scale. First-author manuscript under review at NeurIPS.
Education
University of Illinois Urbana-Champaign — B.S. Computer Science & Economics
2024 — 2027
Statistics minor.
Selected work
Meteorite classification — PyTorch, TensorRT
2023
CNN-based classification with edge-oriented inference for geospatial imagery.
Aegis Notes — Electron, TypeScript
2024
Cross-platform notes with markdown editing and Electron IPC persistence.
Investor sentiment — FastAPI, MongoDB
2023
NLP pipelines and storage for unstructured text and dashboards.
Archive
Earlier roles and side projects. Accurate, but ancillary to the work above.
STEM Enhancement in Earth Science (SEES) — Research Intern, UT Austin Center for Space Research
2022
Bayesian classification and pairing guidelines for meteorite catalog data; analysis of 10k+ historical records in Python, TensorFlow, NumPy. Pairing framework based on physical attributes, geography, and orbital history. Related work presented at the AGU Fall Meeting.
RingCentral — Software Engineering Intern
2022
Cloud communication and video services. Latency-oriented analysis in Python and Tableau, D3 / Tableau dashboards.
Aisera — extended notes
Ontology-driven neural search for retrieval-augmented systems with structured entity hierarchies and dense embeddings. Unsupervised clustering with HAC, UMAP, HDBSCAN to organize enterprise knowledge bases. FAISS ANN indices (IVF-PQ, HNSW) for low-latency search over multi-million-vector corpora.
Investor analysis dashboard — FastAPI, MongoDB, NLP
NLTK and spaCy over investor-facing text; MongoDB schema scaled to hundreds of thousands of rows. Dashboard-facing API tuned for high read volume.
Aegis Notes — Electron, TypeScript, Tailwind, Jotai, TensorFlow.js
Markdown editor with preview; persistence via Electron IPC across platforms. Lightweight tagging and similarity hooks with TensorFlow.js.
Other — smaller projects and writing
UPenn Wharton program (2023). Founded Marine Turtle Foundation, a nonprofit. Built Everything Kindle, a small resale workflow. Earlier writing on education inequality.