Sana Moin

Developed a scalable asynchronous RAG application using FastAPI and Qdrant, deployed on AWSCloud, enabling efficient personalized e-learning experiences for soft skills domain. Built an emotional and speech analysis system with Flask, Celery, Whisper, and LLMs, enhancinguser engagement through advanced feature recognition. Restructured and stabilized backend systems, integrating Kubernetes for container orchestrationand GitLab pipelines for CI/CD to ensure seamless deployment.

Designed and deployed MarFM, an automatic speech recognition(ASR) system for noisy maritime speech, achieving a 20% reduction in word error rate through transformer models, audio preprocessing, and synthetic data generation. Visited a maritime site in Finland to gather customer requirements and implemented distributed training using PyTorch Lightning and Distributed Data Processing (DDP), scaling models across GPUs and delivering near-real-time transcription with a high-performance ASR architecture.

Built an Extraction-Transform-Load (ETL) pipeline for processing hospital data using JavaScript, Java, and SQL, achieving a 40% increase in processing speed. Collaborated with cross-functional teams, including DevOps, databases, and customer-facing teams, to deliver production-level upgrades and enhancements in a scrum environment. Developed a functional testing tool, reducing testing efforts by 80%. Independently resolved 60% of team bugs, earning multiple recognitions for ownership and fast delivery.

Research oriented program integrating concepts from artificial intelligence, machine learning and robotics.

Data Mining, Big Data, Advanced DBMS, Data structures and algorithms