ABOUT ME

Bloomberg
Software Engineering Intern

- Architected and implemented high-performance BatchConsumer API for C++ Kafka library.
- Increased throughput 500x compared to previous offerings (9 msg/s to 5000+ msg/s).
- Ensured system reliability and data consistency with strong ordering guarantees despite unpredictable data schemas and system failures.

New York Structural Biology Center
Machine Learning Intern

- Built and deployed a scalable FastAPI/TorchServe API to host Topaz (cryo-EM particle-picking) machine learning models.
- Optimized model serving for real-time inference and training, integrated logging/monitoring, and resource management.

Tailbox
Machine Learning Intern

- Trained custom embedding and cross-encoder models that improved RAG pipeline results by 14% and reduced latency 50-fold compared to OpenAI models.
- Fine-tuned Qwen2-1.5B SLM to convert HTML text to structured database entries to reduce reliance on private models and save $20/day per city; optimized inference for 19-fold speedup.
- Wrote a robust Requests & Selenium scraper that collects tourist events found online; collected dataset of 18k+ query:activity pairs; used Google OCR to cut costs by 38.8% and cut latency by 27.5%.

Amazon
Software Engineering Intern

- Developed Express.js API to interface custom-trained LLM models and the AlexaEmulator.
- Enabled simulating customer interactions with specific devices through serial numbers/customer IDs.
- Extended project to use Microsoft AutoGen to allow for chain-of-thought reasoning and tool usage.

MIT Computer Science and AI Laboratory (CSAIL)
Machine Learning Researcher

- Co-authored a research paper; accepted by ACM UIST Conference.
- Implemented the software as a plugin on Blender; backend runs CLIP on Django framework.
- Added 1000+ annotated 3D-model datapoints autonomously using Selenium.
- Created website for model using React, Node.js, and Django; provided model-uploading, re-meshing, and stylization in-browser.

Relari Software Engineering Intern

- Implemented REST API using Express.js to dynamically generate tailored company slides by interfacing a proprietary LLM (trained on users' documents) with a PostgreSQL database.
- Engineered document processing pipeline, enabling LLM to extract, categorize, and synthesize key financial data from diverse document formats.

Bausch Advanced Technologies
Software Engineering Intern

- Utilized xTuple API and company database to display part availability and work/purchase order retrieval, dramatically reducing human labor time.
- Implemented SolidWorks 3D viewer directly in-app; includes feature tree, missing parts lookup, and printing.

Bachelor's Degree MIT

Relevant Coursework:

• Computer Science with Python
  Computational Thinking and Data Science
  Fundamentals of Programming
  Low Level Programming with C & Assembly
  Computation Structures
  Operating Systems Engineering
  Systems Design
  Distributed Systems
  Algorithms I
  Algorithms II
  Discrete Math
  Linear Algebra
  Probability
  Machine Learning

Awards:

• Albert G. Hill Award for Community Service

High School Scholar Wesleyan University

High SchoolMorgan School

Valedictorian - 4.4/4.0 GPA