WORK EXPERIENCE

Chief Data Scientist - Persistent - 2013 to present - Goa

Driving strategic initiatives in advanced language models, knowledge graphs (KG), contextual bandits, federated learning (FL), AI for cybersecurity and more. Championed Persistent’s Responsible AI offering that helps drive fast, automated, low-code MLOps pipelines with analysis of model interpretability, explainability, fairness, bias and causal inference. Developed knowledge platform offering that leverages SOTA language models (BERT, BioBERT), graph databases, embeddings and GNNs to enable diverse outcomes from drug discovery, pharmacodynamics to fraud prediction and recommender systems. Building an adaptive enterprise-wide cybersecurity platform to enable network anomaly detection (using auto-encoders), application logs analysis (graph algorithms) and adaptive incident response (contextual bandits).

Principal Architect - Wabtec / GE Transportation - 2013 to 2019 - Bangalore

Championing Innovation in Artificial Intelligence across Transportation domains like – Inspection, Services, Logistics, Mining, Controls, etc. Architecting a Kubernetes-based Platform providing standardised path-to-production for running Physics-based and Machine Learning (ML) models at scale. Driving outcomes like Model Management, AI Workbenches (H2O.ai), ML pipelines (Kubeflow), GPU clusters, auto-scaling of deployments (Serverless, FaaS) and Digital Twins.

Leading the LocoVISION Analytics team on building Deep Learning models for Railway use-cases like – Highlighting Track anomalies, detecting objects like Pedestrians, Signals, Mileposts and monitoring Driver fatigue. Developing the Cloud GPU infrastructure for handling very high volumes of video data (300GB/day/Locomotive) – and spinning up an Elastic Compute environment to run Analytics.

Working closely with Product Management and Commercial teams for Customer engagements in India and SEA region. Driving Thought Leadership in AI through talks at several International Conferences. Hands-on experience in Python, OpenCV, Spark, TensorFlow, Keras, Ionic framework, DC/OS, Docker, Kubernetes, etc. Leading the Indian Railways LocoVISION pilot for video inspection of Railway Track Health.

Chief Software Architect - GE Power - 2007 to 2013 - Norfolk, VA

Developed the multi-generation Technology roadmap for GE Power's Remote Monitoring & Diagnostics (RM&D) centre in Atlanta - that monitors more than 2000 Gas Turbines 24x7 across the world. Simplified the overall System Architecture both at Edge and back-office by incubating many Innovations like standardised Information Model, Common Rules Engine and Orchestration layer. As part of Common rules Engine - evaluated multiple Condition Monitoring models including Fuzzy Logic and Neural Networks - and developed a unified environment based on OSGi. Integrated a shared Information Model for this environment by adopting industry standards like OPC-UA, MIMOSA and Open O&M.

Lead Software Engineer - GE Power - 2004 to 2007 - Norfolk, VA

Developed a Java-based Enterprise portal for deploying Condition Monitoring rules to 1000s of Gas Turbines at remote locations globally. Actively worked with teams to understand analytics using models like Fuzzy Logic and how to deploy them effectively in field. Developed solutions to interface with Plant Software like HMI, Historian and CMMS systems using standards like MIMOSA, OPC and OPC-UA. Involved in many Customer engagements to understand Engineering domain knowledge of Power plants.

Research Engineer - GE Global Research - 2000 to 2004 - Bangalore

Developed a Knowledge-based Engineering (KBE) solution for Transportation CAD systems. Leveraged System-level thinking to co-relate high level requirements and flow them down to individual component geometry changes. Developed solutions using Unigraphics and Knowledge Fusion to capture Design Knowledge into the system - through multiple interviews with Designers understanding their thought process.