Skip to content

Portfolio

Projects

Research and production work spanning ML, full-stack systems, and cloud infrastructure. Each project includes a detailed write-up.

Research
12.6M

Multiplayer Frames

Solaris — Multiplayer Video World Model in Minecraft

First multiplayer video world model generating consistent first-person observations for two players simultaneously, trained on 12.6M frames of coordinated Minecraft gameplay. Published on arXiv, NYU.

JAXDiffusion TransformerMultiplayerMinecraftComputer Vision
Read more
Full-Stack
v2.0

Shipped

Teserax.io — Graph-Based AI Thinking Tool

A dual-lane, chat-first exploration tool that transforms linear LLM chat into a visual, non-linear graph canvas with AI orchestration, crosslink reasoning, multi-model BYOK support, and cloud persistence — shipped v2.0 with 64 issues closed across 5 development phases.

ReactReact FlowHonoZustandSupabase
Read more
Full-Stack
96.57%

Accuracy

Cloud NLP Classification on GCP

Production-ready multi-model text classification service with zero-downtime model switching, deployed on GCP Compute Engine. DistilBERT trained to 96.57% accuracy on a 24,783-sample dataset; 326+ test suite at 100% pass rate.

NLPDistilBERTFastAPIDockerGCP
Read more
ML / AI
91.3%

Accuracy

MRI Brain Tumor Detection & Segmentation thumbnail

MRI Brain Tumor Detection & Segmentation

Multimodal MRI classification and segmentation model trained on BraTS dataset with shared encoder, achieving 91.3% accuracy and 97.1% sensitivity.

PyTorchComputer VisionMedical ImagingFastAPIBraTS
Read more
Research
73.21%

Token Accuracy

PICO-LLM Research Pipeline

Modular LLM research pipeline (NYU CSCI-GA 2565) for training and evaluating K-Gram MLP, LSTM, and KV-cache Transformer architectures with 22+ experiment configs and rigorous cross-run analysis.

PyTorchLLMTransformersLSTMKV-Cache
Read more
ML / AI
1st Prize

CVS GenAI Hackathon

AI-Driven Image-to-Alt-Text CI Automation

Hugging Face transformer-powered image alt-text compliance system at CVS Health. Won 1st Prize at the CVS GenAI Hackathon (Aug 2023), then productionized for enterprise rollout — boosted accessibility conformance by 35% and reduced downstream defects by 15%.

Hugging FaceComputer VisionTransformersCI/CDGitHub Actions
Read more
ML / AI
20%

Faster Resolution

RAG Support Assistant

Internal RAG-powered support assistant with Slack integration, OpenAI, and ChromaDB. Hosted on CVS Enterprise GCP via Terraform + ArgoCD; reduced manual ticket resolution time by 20%.

RAGOpenAIChromaDBSlackTerraform
Read more
Open Source
ACM

ASSETS 2023

Testaro — Web Accessibility Testing

Multi-tool web accessibility compliance evaluation and unified reporting pipeline. Presented at ACM SIGACCESS ASSETS 2023 representing CVS Health.

AccessibilityNode.jsCI/CDPlaywrightACM
Read more
ML / AI
1st Prize

Aug 2023

CVS GenAI Hackathon — Alt-Text Compliance

Awarded 1st Prize at the CVS GenAI Hackathon (Aug 2023) for a proof-of-concept ML transformer application that generates and compares image alt-text and tests it for accessibility compliance. Productionized into the CVS image-to-alt-text CI pipeline.

Hugging FaceTransformersHackathonAccessibilityComputer Vision
Read more
Open Source
1st Place

Galactic Impact

Atlas Overlay — NASA SpaceApps Challenge

Construction planning tool leveraging NASA satellite imagery, terrain segmentation, and CV-based analysis. Awarded 1st Place — Galactic Impact at NASA SpaceApps Cleveland 2021 and Best Trine Research Project (Engineering Multi-Departmental) at Trine University's STEM Symposium.

Computer VisionGISArcGISAutodesk RevitSatellite Imagery
Read more
ML / AI
Capstone

Trine Senior Design

Automatic Nerf Remote Sentry

Senior capstone — fully autonomous Nerf turret with OpenCV-based real-time target tracking, motion prediction, and gyroscopic aiming on a Raspberry Pi 4. Trine University Robotics Senior Design (Mar 2021).

OpenCVComputer VisionRaspberry PiRoboticsPython
Read more