moffat adam

I’m a computer science undergraduate focused on building real-time software systems, modular pipelines, and deployed full-stack applications. I bring a strong foundation in OOP, data structures, and compute-heavy workload optimization to deliver production-grade outcomes.

In my Vision Module project, I architected a modular real-time perception pipeline (perception → feature extraction → intent inference → execution) in Python. I sustained 60+ FPS with <100 ms latency, achieved >95% gesture accuracy with <3% false positives, and built an event-driven automation engine supporting 10+ system actions within a <500 MB memory footprint.

For Brain Tumour Detection, I designed and deployed a full-stack web application using a Flask backend and Streamlit frontend. I improved CNN accuracy from 85% to 95% through advanced data augmentation and reduced average prediction response time by 30%, serving 150+ unique users and leading processing/classification of 200+ MRI scans.

With TerraSight, I’m developing a modular multimodal LLM pipeline for automated land-use classification across 1,000+ ISRO satellite images, reaching ~85% accuracy while reducing inference latency by ~30% through batching and optimized preprocessing. I also led my team in Deloitte Hacksplosion to advance to Round 3 among 22,000+ participants, and I’m pursuing credentials in SAP and MongoDB.