Applied AI Engineer · Data Infrastructure · Machine Learning
I am a Master's student in Computer Engineering at NYU and Research Assistant at NYU Grossman School of Medicine, where I fine-tune medical vision-language models (MedGemma, MedSigLIP) on 50,000+ skin lesion images for smartphone-based mpox detection — 92% accuracy — advised by Dr. Megan Coffee.
At mLab, I work on data labeling infrastructure for healthcare ML using passive network traffic analysis to detect Alzheimer's disease, advised by Dr. Danny Huang.
I build products, ship fast, and iterate with users. Currently building Nidra — a training optimization tool aggregating Whoop, Apple Health, and network activity data. Obsessed with extracting maximum performance in software engineering and athletics alike.
Built Agentic AI workflows for 5 clinical trial personas, reducing automation time by 45%. Developed BERT-based chatbot achieving 92% query resolution and 60% reduction in support workload.
Processed 500GB+ daily financial datasets for 54 F500 clients, enhancing anomaly detection by 22%. Built dashboards that increased client revenue by 17%.
Built data pipelines supporting 40% volume increases with sub-second response times. Migrated SAS to PySpark, reducing execution time by 13% and improving accuracy by 47%.
Built IoT RFID security system reducing Loss Time Incidents by 83 man hours. Optimized test cell power consumption by 34% using data mining algorithms.
Photo album with natural language search using AWS Rekognition and Lex. Serverless architecture — Lambda, S3, OpenSearch.
AI dining concierge with NLP. Serverless architecture integrating Yelp API with 5,000+ restaurant entries in OpenSearch.
Library management system with 70+ REST API endpoints supporting 10,000+ books across 4 user roles. JWT auth, room booking, multi-payment processing.
Cycle-accurate RISC-V processor simulator for performance analysis. Complete RV32I instruction set with multi-stage pipeline.
Wearable for Parkinson's patients detecting tremors (3–5 Hz) and dyskinesia (5–7 Hz) via real-time FFT on accelerometer data.
I write about software engineering, product development, and the philosophy of building. Read my essays ↗.
Interested in collaborating, discussing ideas, or chatting about software engineering and training optimization? I'd love to hear from you.