Avinaba Roy Email & Phone Number

Sydney, New South Wales, Australia

Sydney, New South Wales, Australia

Sydney, New South Wales, Australia

Sydney, New South Wales, Australia

- Development and deployment of a highly complex set of Big-Data Analytics and Machine Learning pipelines whose primary objective is to classify and manage the client's HFC (Hybrid Fibre Co-Axial) and Fixed Wireless (FW) network degradation incidents proactively and enable end-to-end traceability of the genesis of the network fault.- Development and deployment of a highly complex “Stitched” data model comprising of Network Telemetry data, Network Topology Data, Geo-spatial Data, Satellite Imagery and Weather data (Bureau of Meteorology) which is capable of powering multiple use-cases with respect to the client's HFC (Hybrid Fibre Co-Axial) and Fixed Wireless (FW) infrastructure.- Development and deployment of “Layered” Semi-supervised ML pipeline(s) involving a combination of clustering and classification machine learning algorithms coupled with synthetic oversampling data augmentation strategies for the class imbalance resolution, with False-Discovery-Rate (FDR) as the core evaluation metric.- Investigated "end-to-end" remediation workflows with respect to HFC and Fixed Wireless services and progressively incorporated cognitive automation based value levers leading to significant operating expenses (OPEX) cost reductions.

Sydney, New South Wales, Australia

- Developed a complex Rule(s) based data model involving significant levels of data augmentation and derived attribute(s) formulation associated with Issue-based and Return-based delivery activities.- Exploratory Data Analysis (EDA) coupled with correlation analysis and pattern extraction in order to establish statistical linear relationships with respect to vehicular deployment aspects such as capacity and nature of cargo versus the transport and labour costs involved.- ML powered insights with respect to logistics cost optimisation across routes, vehicular demand forecasting, Issue/Returns specific delivery activities and Labour-Hours optimisation.- Leveraged patterns and forecasting algorithms coupled with additional business logic (such as sensitivity/volatility of the cargo) to drive value in the form of vehicular deployment optimisation and efficient grouped deliveries strategy.

Sydney, New South Wales, Australia

- Designed and developed complex Data Models, Analytics Insights in conjunction with sophisticated Data Mining modules in the form of custom Web Crawlers which were deployed to gather and process sensor data pertaining to legacy telecom equipment that was part of the client's cable network infrastructure.- Developed and deployed Rapid Prototypes in order to establish a solid foundation of the solution design decisions.- Developed and deployed production-grade code and cross-platform (Windows/Mac/Linux) tools using a complex combination of technology stack maintaining full operational integrity with respect to the client’s operating environment.- Identified and resolved 'Pre' and 'Post' production issues and/or blockers consequently ensuring that stakeholder expectations are maintained at all times.

Sydney, New South Wales, Australia

- Spearheaded the successful delivery of a Major “End-to-End” Enterprise Natural Language Processing (NLP) Solution for one of the Top 4 Banks in Australia as part of their Business Process Transformation Initiatives.- Development and deployment of an Intent Classification Engine along with an Entity Extraction Engine with the sole objective of performing a feasibility analysis between a completely ”Home-Grown” fully customisable open-source solution versus commercial “Off-the-Shelf” products.- Designed and developed production grade Natural Language Processing (NLP) solutions in the form of Virtual Agents in-line with scalable enterprise architecture design principles that met the client’s requirements.- Led a team of developers in order to ensure module delivery within the designated time-frame(s).- Developed and deployed rapid Proof of Concepts (POCs) that establishes the core capabilities and constraints of the client’s operating platform.- Phase-by-Phase Monetary impact assessment of the solution as it evolved and scaled across different business verticals within the scope of the target Business Process Transformation.

Sydney, New South Wales, Australia

- Developed analysis and insights with respect to the mechanics and principles governing Transaction Categorisation in an on- premise custom Big-Data Environment consequently formulating Rules and/or Business Logic with respect to Transaction Categorisation.- Development and deployment of a hybrid solution involving a complex Rule-based engine and Natural Language Classifiers (NLCs) applicable to the merchant transaction text data powered by advanced feature extraction capabilities which were ultimately used to train and assess different multi-class classification algorithms keeping in mind high-dimensional sparse nature of the data.- Performed experimental analysis with respect to short-text classification with Deep Neural Networks (Convolutional Neural Networks CNNs) powered by pre-trained word embedding(s) applicable to the Open-Banking Use-case which involved purely short-text customer transaction statements.- Effective stakeholder(s) engagement with respect to a communicating a 'Pre' and 'Post' Analysis snapshot of the evolving requirements complexity of the problem statement.

Sydney, New South Wales, Australia

- Developed and deployed customer data linkage and data enrichment in order to generate data models that can be leveraged by the different business units.- Generated nearly 26000 derived/calculated attributes for 15 million records as part of the Single Customer View. Solution was in the form of a hybrid open-source and cloud-based technology stack which also involved integrations with the customer’s existing on-premise data platform.- Customer Demographics, Transaction History, Event Attendance, Marketing Responsiveness were some of the variables calculated as part of the feature engineering process.- Deployed Dimensionality Reduction strategies coupled with clustering algorithms in the context of the different business units in order to extract relationships/trends between the applicable classes (such as popularity of specific event types for a specific subset of common customers)

Canberra, Australia

As part of the Altair Australia Team I was involved with the Bureau of Meteorology, specifically working on the "Cylc-PBS Integration" Project. "Cylc" is a workflow engine for cycling systems - it orchestrates complex distributed suites of interdependent cycling tasks that may continue indefinitely and PBS Professional is a fast, powerful workload manager designed to improve productivity, optimise utilisation & efficiency, and simplify administration for HPC clusters, clouds and supercomputers. Therefore, the primary objective of this project is to ensure the seamless integration of the "Cylc" Workflow Engine and PBS Professional. With the exception of the aforementioned projects, I have also been involved with the following tasks:-SGE to PBS Script conversion and compatibility, primarily for job submission purposes coupled with support for multiple Linux shells.-Script Validation framework (Python Based).-Extensive investigation of the CRAY Resource Utilisation Reporting(RUR) module, in order to accommodate HPC accounting metrics in conjunction with PBS Professional via advanced PBS Hooks and Python scripts.

Bangalore , Karnataka , India

Was a member of the PBS Professional Core development team. PBS Professional is a state-of-the-art workload manager and job scheduler for high-performance computing (HPC) environments.Some of the responsibilities(tasks) that were executed by me are:- Developed and implemented IPC (Inter-process communication) protocols and advanced networking modules customised for the National Computational Infrastructure(NCI) Australia.- Developed high speed data parsers in C and Python for various modules in conjunction with PBS Professional (Workload Manager/Job Scheduler).- Developed advanced usability feature set for PBS Professional (Linux based).- Developed and implemented multiple Graphical User Interfaces complementing PBS Professional using Python (PyGTK and wxPython).- Developed critical enhancements and fixes as part of future releases for PBS Professional.- Performed application enhancements on select software suites that revolve around PBS Professional.

Bangalore , Karnataka, India

Primary objective was to understand PBS (Portable Batch System) which is a state- of-the-art workload manager and job scheduler, and develop a GUI (Graphic User Interface) that provides an easy-to-use interface to submit jobs and monitor them along with other cluster management tools. Also in order to support enterprise energy conservation initiatives, "Green Provisioning" power management module has to be integrated in the GUI which consists of advanced scripts and hardware factors that help reduce energy usage by almost 30 percent. The tasks that were performed by me are:- Developed suitable tools in Python 2.6 and its wrapper class wxPython (GUI API toolkit) as front-end and utilising PBS programmable APIs written in C as back-end with administrator suitable features and functionality.- Develop PBS Hooks, which are python scripts that can submit a job, alter the set of conditions governing the submission of jobs and integrate them into the PBS Professional environment.- Performed field testing of the GUI on various clusters and systems which run PBS Professional and generate linux binaries and windows distributable accordingly.

Gunsan, Jeollabuk-Do, Korea

The primary objective was to understand the advanced concepts of CAN (Controller Area Network) and its higher layer protocol J1939 along with its design and implementation. In networking ECUs (Electronic Control Unit) in heavy-duty vehicles, J1939 protocol plays a key role. J1939 networks are primarily used in powertrain and chassis components. Some of the tasks performed by me are:- Performed design ,implementation and simulation of a simple J1939 CAN network using "Vector Informatik CANalyzer".- Assessment of Real time J1939 Data of the Tata Daewoo Prima module and data manipulation with CAPL ( CAN Access Programming Language) which is an event-based programming language.

Jamshedpur Area, India

Worked on a project titled "Job Card Management System" which involved the replacement of paper-based job card(s) of vehicular parts with a software system.Objective was to develop a Job Card management system for automotive assembly equipment as a replacement for the manually-operated paper- based card system. Image data of the job card is processed by a scanner that reads the barcode and converts it into xml file and also the image on the job card into pdf file. User can view the details of "Job Card" and can discard/save them as per the requirement. The xml file detail is stored in the database and a web application is developed to view the different reports and graphs and user settings. The tasks that were performed by are:- Worked with Oracle 10g Database which served as the project backend- Developed UI (User Interface) elements in ASP.NET using Visual C# which served as the project front-end

Bengaluru Area, India

The primary focus of the internship was to investigate the usage and application of the Siemens Teamcenter Product suite, specifically "Tecnomatix Robcad". The tasks that were performed by me are:- Performed modelling, simulation and offline programming of Robotic Spot Welding Application using Teamcenter product-suite "Tecnomatix Robcad".- Performed accurate simulation and synchronisation of multiple robots and mechanisms.

Jamshedpur Area, India

Worked on a project titled " Study of FANUC M900i series of Robots in Automotive Manufacturing " which involved a detailed analysis of its capabilities from an automotive assembly equipment standpoint along with an in-depth study of its programming interface. Some of the tasks that were performed by me are:- Performed a detailed study of FANUC M900i series of robots which is a six- axis, electric servo-driven family of robots designed for a variety of manufacturing and systems processes.- Performed a detailed analysis of the programming interface using the R-J3iC controller.

Master Of Computing, Artificial Intelligence And Professional Computing

Bachelor Of Engineering (B.E.), Electrical And Electronics Engineering

I.C.S.E (10Th ) And I.S.C (12Th )