Abhilash Sadasivan

Built a state-aware anomaly detection system on multivariate sensor streams with real-time inference on edge hardware. Designed an attention-augmented LSTM Autoencoder that adapts to arbitrary sensor configurations, enabling deployment across sites without re-architecting per site Developed the full ML pipeline: ingestion, preprocessing, training, validation, live inference Built a Dash + SQLite operator dashboard for on-prem real-time anomaly visualization, removing cloud dependency Profiled CPU/GPU/m

Computer-vision inspection pipeline work and AI hardware evaluation for production deployment. Replaced EfficientNet with YOLO in the inspection pipeline, cutting inference latency from ~400 ms to ~25 ms. Evaluated industrial AI hardware candidates, scoring on inference performance, integration, and procurement fit Resolved vendor ML environment issues to enable hardware testing

Developed comprehensive web application for offline banking and customer information management, serving as the core infrastructure for a major Indian insurance provider. Resolved critical system vulnerabilities by spearheading an upgrade to the latest versions. Meticulously documented the entire process to enable future upgrades, after prior stalled attempts over two years faced complications. Lead team during difficult transition phases. Leveraged existing knowledge of Java, SQL, HTML, CSS and Eclips

Key subjects: Artificial Intelligence, Object Oriented Programming, Discrete Computational Structures, Automata Languages and Computations, Data structures and Algorithms, Database Management Systems, Advanced Computer Networks, Advanced Architecture and parallel Processing

Key subjects: Artificial Intelligence, Object Oriented Programming, Discrete Computational Structures, Automata Languages and Computations, Data structures and Algorithms, Database Management Systems, Advanced Computer Networks, Advanced Architecture and parallel Processing