Experience

  1. Software(ML) Engineering Intern

    Metaphor Data Inc.
    • Developed comprehensive testing framework for 3 LLM and LangChain pipelines (search, automated descriptions, Slack bots). Identified 2 critical bugs and improved performance by 20%. Leveraged MongoDB, TypeScript, Jest, CI/CD, AWS, Terraform, and Langchain for efficient retrieval and robust system architecture.
    • Pioneered first CLI tool and 4 new API endpoints for 3 batch transactions across 5 data asset types, increasing efficiency by 10x. Implemented using GraphQL, Python, Pydantic, and TypeScript to ensure flexible querying and type-safe integration.
  2. Undergraduate Researcher

    Focal Lab - UIUC
    • Developed novel method on Certification of Reasoning using Knowledge Graphs for LLMs, accepted at SetLLM@ICLR
  3. Head of AI/ML, Tech Lead, SWE

    Disruption Lab - UIUC
    • Led and collaborated across 10 AI project teams with 100+ students, achieving a 91% project success rate for clients including AMD. Projects spanned:

      1. LLM-guided web crawling: Implemented scalable data collection automated pipeline with custom scrappers and Mistral-7b for intelligent navigation.
      2. Virtual classrooms: Developed a virtual classroom using Unity to support interactive teaching. Developed chatbots with LlamaIndex and llama-2 for answering students questions using avatars.
      3. Malware detection: Collected Novel Dataset for AMD with hardware counters on systems with and without malware. Implemented a baseline LSTM for identifying randomware with 70% accuracy.
    • Implemented distributed training on AWS SageMaker using PyTorch DDP, reducing LLM fine-tuning time.

  4. Undergraduate Researcher

    Kreiman Lab - Harvard Medical School
    • Introduced novel dataset on humor detection for Vision and Vision-Language models
    • Gained experience with DeepSpeed and Huggingface accelerate for distributed model training
    • Preprint on ArXiV
  5. SPIN Intern

    National Center for Supercomputing Applications - UIUC
    • Presented poster on using Diffusion Models and Inpainting for Parkinson’s disease simulation on facial features
    • Achieved 6% improvement in simulation accuracy
  6. Software Developer

    UIUC TheorieLearn
    • Developed 12 educational resources using JavaScript, HTML, Python for aiding student learning online for the Algorithms Class at UIUC.
    • Helped Fix 5 known issues to improve student feedback.

Education

  1. BS Mathematics&Computer Science

    University of Illinois Urbana-Champaign

    GPA: 3.98/4.0

    Courses included:

    • Trustworthy Machine Learning
    • Embodied Natural Language Processing
    • Compiler Construction
    • System Programming
    • Numerical Analysis(Methods)
    • IoT & Wireless Sensing
    • Formal Methods
Selection of Skills
Technical Skills
Python
C++
PyTorch
Machine Learning
Formal Methods
Compilers
Wireless Sensing