Hi, I'm Dhira.
Right now I'm making vision-language models run fast on small hardware.

As of September 2026

I'm a computer science and biology student at the University of Maryland. Lately that means fine-tuning vision-language models to run on a 15-watt board strapped to a backpack, hardening LLM guardrails against prompt injection, and shipping the software a real club relies on every day.

Now
Software Engineer, AI Team at UMD DIT
Last summer
SWE Intern, GDIT
Studying
CS (Machine Learning) & Biology, UMD '28

Current projects

An NVIDIA Jetson Orin Nano developer kit on a desk next to a bagel of roughly the same size
Bagel for scale.

On-device assistive vision for blind and low-vision users

Veris

The software side of a privacy-first wearable that describes the world out loud from a fine-tuned vision-language model running entirely on a Jetson Orin Nano. No image ever leaves the device.

  • PyTorch
  • QLoRA
  • llama.cpp
  • whisper.cpp
  • FastAPI
  • Docker
The Beanwatch analytics page showing spend trends, cost per event, and consumption forecasts

Operations platform for the UMD Coffee Club

Beanwatch

Inventory, event, and delivery management that tracks every bag of beans from delivery to the bottom of the cup. In daily production use by the club for over a year.

  • React 19
  • Supabase
  • PostgreSQL
  • Row Level Security
  • Edge Functions
  • Vercel

Experience

  1. May 2026 – Aug 2026Herndon, VA

    Software Engineering Intern

    GDIT · General Dynamics Information Technology, Intelligence & Homeland Security

    • Built a machine-learning classifier that detects prompt-injection and jailbreak attacks, reaching 96% recall on 543 unseen attacks.
    • Found and repaired a 46% label-corruption problem in the training data through iterative retraining.
    • Added uncertainty-based guardrails that flag unfamiliar attacks, and wrote a proposal that was adopted for the next generation of internal guardrail systems.
    • Shipped a Flask web app demonstrating the full detection pipeline to stakeholders.
    • Python
    • scikit-learn
    • PyTorch
    • Flask
    • LLM security
  2. Feb 2026 – PresentCollege Park, MD

    Software Engineer, AI Team

    University of Maryland, Division of Information Technology (DIT)

    • Built a one-click, end-to-end Playwright QA pipeline for a document-review product, moving the release cadence from bi-weekly to weekly.
    • Created an automated bug-repair pipeline that feeds Playwright failure output to LLMs to propose fixes to production code.
    • Developing an onboarding chatbot with Jira and Atlassian connectors to answer new-hire questions.
    • TypeScript
    • Playwright
    • LLM tooling
    • Atlassian APIs
  3. May 2023 – Aug 2023Annapolis Junction, MD

    Software Engineering Intern

    Praxis Engineering

    • Developed an NLP application for referencing past corporate projects, cutting research and summarization time by 65%.
    • Reverse-engineered router firmware in Ghidra to identify vulnerabilities, building C and Python tooling for the analysis.
    • Python
    • C
    • NLP
    • Ghidra

Computer science on one side, biology on the other

I'm pursuing two degrees at UMD: computer science on the machine learning track, and biological sciences with a bioinformatics focus. The pairing is deliberate. The ML problems I find most interesting are the ones with a physical constraint attached, whether that's a power budget on an edge device, a person who needs an answer in under four seconds, or a dataset that came from the messy real world and needs to be cleaned before anything learns from it.

Outside of class I co-founded Veris, an assistive-vision wearable backed by UMD's AI institute, and I run the analytics team behind the Coffee Club's operations platform. Before college I earned Eagle Scout and led a troop of over a hundred people, which taught me more about shipping things with a team than any course has.

I'm looking for summer 2027 internships and co-op placements in machine learning engineering, applied ML research, or backend systems. If that sounds like your team, say hello.

What I work with

Languages

  • Python
  • TypeScript / JavaScript
  • Java
  • C
  • Rust
  • SQL
  • Bash
  • x86 Assembly

Machine learning

  • PyTorch
  • Hugging Face Transformers & PEFT
  • QLoRA fine-tuning
  • llama.cpp / GGUF quantization
  • LLM distillation & eval design
  • whisper.cpp
  • scikit-learn
  • vLLM / TensorRT benchmarking

Systems & tools

  • FastAPI
  • Flask
  • React
  • Node
  • Playwright
  • Docker
  • Supabase / PostgreSQL
  • NVIDIA Jetson
  • Git
  • Azure
  • Figma

Leadership & activities

  • Eagle Scout & Senior Patrol Leader2016 – 2024

    Led a troop of 120+ members and organized more than 1,000 hours of community service.

  • UMD Coffee Club Executive Board2024 – Present

    Founded the analytics team and built the club's operations platform.

  • UMD Terps Racing (EV), electrical sub-team2024 – 2025

    Designed thermistor sensor systems for the electric race car.

When I'm not writing code

The things I do for fun, and the odd ways they feed back into the work.

  • Food

    Eating is the hobby. I keep an album of the good ones, most of it from two weeks in Japan and one very long dinner in Osaka.

    See the album
  • Competitive Magic: The Gathering

    I play cEDH, the competitive end of Commander.

    Current Tournament List
  • Singing

    I competed from a young age and sang seriously through school. These days it is just for me, which is how I like it.

  • Mechanical keyboards

    Switches, keycaps, layouts, firmware. I am always chasing a slightly better typing feel, and I will talk about it longer than you want.

Let's build something useful.

The fastest way to reach me is email. I read everything.