1) Developed and deployed a Windows-native, multithreaded imaging sync service (PUB/SUB) across 15+ clinics, automating cloud uploads every 30 seconds for high-volume dental imaging workflows.

2) Optimized performance by reducing memory usage 80% with a fixed-size min-heap approach vs full-directory scans, maintaining <3s upload latency and 99.9% uptime with auto-restart watchdogs and remote logging.

3) Designed scalable ingestion modes — Just-in-Time (50–70 files/min) for active cases and Historic (TB-scale) for archives — using throttled memory strategies and bounded queues.

4) Integrated AWS cloud services (S3, Lambda, SQS) into a fully HIPAA-compliant pipeline, handling 100K+ imaging files/month securely and reliably.

5) Delivered 500+ AI-generated DICOM reports/month by integrating diagnostic AI APIs with PACS servers via DICOM SCP/SCU protocols.

6) Designed and implemented a voice-controlled Perio Charting UI in React, reducing clinician data entry time by 60% while achieving <200ms latency for real-time interactions.

Teaching and Mentoring, Algorithms, Logic, Problem-Solving, Communication, Time Management, Evaluation and Assessment, Python.

Teaching Assistant for Computer Science 1, Introduction to logic, Introduction to Algorithms.

Conducted labs sessions, help sessions for students and graded exam papers and homework assignments.

Federated Learning, Unlearning Methods, Grid Search, PyTorch, Matplotlib, Data Analysis, Model Evaluation, and Customization.

1)Learned about Federated unlearning from the team at IBM.

2)Used methods such as grid search and tools such as Matplotlib and pytorch to evaluate the proposed method of unlearning and customize it for different datasets.

3) Implemented a new method of evaluation of federated unlearning.

I worked on improving many of the APIs in the source code that dealt with video sharing and recommending. using Node.js and SQL