Project: Service-Affecting Failure Root Cause AnalysisA solution to identify the root-cause of service affecting failures in Urban Transit Systems for effective troubleshooting.• As Data Scientist – Performed Causal Discovery on data-logs to obtain a relationship graph, revealing the underlying causal structure between events and failures. – Developed a Causal model and applied Interventional and Counterfactual techniques to identify root causes of failure.• As Data Engineer – Created and deployed end-end Data pipeline for Data Ingestion, Data Processing (Cleaning, Transformation), and Data Visualization. – Created and deployed a web-based pedagogical tool to share results with stakeholders.

Project: Log AnalyticsAn Analytical solution for detecting undesirable system behaviour from datalogs.• As Data Scientist – Worked with system experts to define the Health Indicators to detect the undesirable behaviours. Design rules to detect erroneous patterns in the logs – Developed PowerBI dashboards application to help troubleshooters identify unhealthy patterns appearing in data-logs and efficiently navigate through them. Thus reducing the troubleshooting time by 80%• As Data Engineer – Developed analytics modules for computing complex health indicators using Big-Data technology – Automated the end-to-end data pipeline (Data Injection, Data Processing (Cleaning, Transformation)) with Apache AirflowProject: Incident DetectionA Data-driven troubleshooting solution to detect unhealthy trains from a fleet that needs maintenance.• Developed a statistical model to detect faulty trains from the event datalogs generated by train.• Packaged solution as executable and deployed in Alstom’s Fleet management system.Project: Anomaly detection in Odometry systemA solution to detect anomalies in the train’s odometry system to ensure efficient and effective maintenance.• Developed algorithm using Autoencoder architecture that detected anomalies with high Recall (over 99%)Project: Ebicab Developed two unsupervised anomaly detection methods based on Autoencoder and textual transformer that detects the variable deviation, explaining the presence of a train software failure in time-series data coming from train's ATC subsystem. Conducted a comparative analysis of both methods and documented the findings in a technical report. The solution is able to identify failures not detected by rule based approach.

Project: RailSketch (Aug 2017 - Jan 2020)- Developed RailSketch, a graphical application that allows users to create animated videos for demonstrating signaling principles.- Enabled users to create track layouts by adding tracks, signals, switches, and other field objects, and to animate train movements and signal state changes.- Implemented collaborative functionality, allowing geographically separated users to work simultaneously on the same track plan for effective signaling demonstrations.- Facilitated early detection of design issues by system engineers during the signaling system design phase.Project: M8-PIU (Dec 2018 - Sep 2019)- Developed the Portable Interface Unit (PIU), a diagnostic application for the ATC module of a train car.- Enabled users to download and view datalogs, event faults, and snapshots for enhanced diagnostics.- Facilitated interaction with various sub-modules such as DBAU and TestBoard for comprehensive system diagnostics.- Implemented two connection modes: Direct mode for serial connections to the subsystem, and Remote mode for connecting via Ethernet using TCP/IP protocol through the Central Diagnostics System (CDS).

Project: GraphSim (Jan 2017 - Dec 2018)- Developed GraphSim, a simulator for iVPI-based interlocking systems, replicating key functionalities such as logic evaluation, sensor input processing, output relay activation, and parameter exchange between vital and non-vital subsystems.- Enhanced the existing simulator with redundancy simulation capabilities, including failover mechanisms, hot standby, and synchronized variable states for improved reliability.Project: FOS (Sep 2015 - Jan 2017)- Developed a TCP-based communication protocol for seamless data exchange between simulation software and signaling hardware systems.- Created a simulator to model and simulate way-side signaling devices, enhancing system testing and validation.- Designed and implemented Automated Factory Acceptance Testing (FAT) processes for interlocking systems, improving efficiency and reducing manual intervention.- Developed interfaces to enable communication between hardware systems operating on different protocols, ensuring robust integration with automatic testing software and hardware subsystems.- Implemented a WCF service interface to facilitate data access within SCADA's OPC UA server.

Development of next generation cancer diagnostics using smart polymer technology- Analyze the market trends for developing next generation cancer diagnosticsusing smart polymer technology- Analyze the needs of consumer to improve the product

Development of state estimation algorithms for estimating the underactuated degrees of freedom in humanoid robots- Development of dynamic models of humanoid robot "Toro"- Design of non linear Kalman filtering algorithms for state estimation- Implementation of algorithms in Matlab/Simulink

Project: Automation of System under test- Programmed solar inverter test specifications using dSPACE AutomationDesk software.- Designed and implemented test cases for solar power inverters.- Established communication between various test equipment using multiple communication protocols, including GPIB, Ethernet, and RS-232.- Commissioned and networked various equipment utilizing different communication protocols.- Designed and implemented CRC code generation algorithms in MATLAB/Simulink for testing microcontrollers in a Hardware-in-the-Loop (HIL) system.- Developed a Human-Machine Interface (HMI) module for solar inverters.- Evaluated results and prepared comprehensive documentation