• Developed and configured custom feeds and reports for an Absence Management project, utilizing SQL, SSIS, and SSRS to meet client requirements over the reporting server.
• Engaged in the maintenance and troubleshooting of SSIS packages, serving as an SQL developer to ensure system integrity and performance.
• Supported the transaction layer of an ASP.NET application, effectively managing a hub-spoke model database on the Transaction server.
• Ensured seamless data integration flow from the user interface to the transaction server, and from the transaction server to the reporting server, maintaining data consistency and accuracy across layers.
• Deployed SSIS packages for efficient data transfer between the transaction layer and the reporting layer, optimizing data accessibility and report generation.

• Developed and analyzed ETL processes using SAS Studio and MSBI tools (SSIS, SSRS, PowerBI) for an airlines analytics project, streamlining data integration, building pipelines and reporting.
• Engineered data integration solutions with SSIS, SSRS, and SAS for enterprise and customer data warehousing, improving data management and analytics capabilities.
• Managed and maintained over 200 ETL flows across Azure Data Factory, SQL Server Agent, and SAS Viya, ensuring efficient data processing and reliability.
• Directed the successful upgrade of the Enterprise Data Warehouse, transitioning and decommissioning SSIS and SQL pipelines from the 2008 to the 2016 version, enhancing system performance and future readiness.
• Monitored Data Warehouses’ ETL flows, resolved repetitive issues by root cause analysis, and optimized server loads through strategic SQL job scheduling, ensuring high data integrity and system efficiency.

• Designing and developing custom feeds and reports for clients.
• Maintaining and troubleshooting the SSIS, SSRS, and SQL packages/ components for any issue.

• Analyzing the SAS Studio (SAS Data Integration Studio, Enterprise Guide) ETL processes and developing the new ETL products in MSBI (SSIS, SSRS, PowerBI) and SQL server for the data analytics team.
• Review the business requirement of ETL flows and design the new interfaces as per the architecture.

Environmental setup for windows from 2008 to 2016 and migration of SSIS packages by re-confuguring/ upgrading the components from SQL 2008R2 to SQL 2016 and testing their compatibility.

Monitoring, the ETL flows for the Datawarehouse deployed in the SQL Server agent job, Creating the maintenance changes through ServiceNow, fixing the repetitive problems by analyzing the root causes and reporting the application status, and scheduling the SQL jobs to maintain the load on the server.

Tech Stack: Python, Pytorch

Project Description:

• Developed and evaluated an interaction-aware text-to-image diffusion model by integrating a pluggable module (InteractDiffusion) into Stable Diffusion v1.4, enabling fine-grained human-object interaction control in generated images using HICO-DET dataset.
• Implemented interaction conditioning using a novel transformer-based module incorporating tokenization, spatial embeddings, and role-aware attention to guide image generation from HOI triplets (subject-action-object).
• Trained and benchmarked the model with FID, KID, and HOI detection mAP metrics, achieving improved interaction fidelity and image quality over baselines like GLIGEN and Stable Diffusion.
• Conducted ablation studies to assess the contribution of each interaction module component, showing progressive improvement in controllability and realism of synthesized images.

Tech Stack: ATP (Alternative Transients Program), Power System Studies, Power System Stability, Transient Stability, Modeling and Simulation

Project Description:

• The large-scale integration of wind generation into the grid has led to increased reactive power requirements and associated stability challenges. This project examines the “CREZ Reactive Compensation Study,” which highlights the effects of Series Capacitance and Shunt Reactive Compensation.
• The study identifies potential challenges in implementing reactive compensation solutions and proposes mitigation strategies. A key issue identified was the potential for sub synchronous interactions, including sub-synchronous resonance (SSR) and sub-synchronous control interaction (SSCI), which can destabilize the system and cause mechanical damage to turbine-generator shafts.
• To investigate the SSR effect, the IEEE Second Benchmark System for SSR Studies was utilized, and several case studies were conducted. Alternative Transients Program (ATP) simulations analysed system behavior under varying conditions, including blocking capacitor values, series compensation levels, fault timings, and a torque amplification study.
• The results demonstrated the necessity of transient simulation for ensuring system stability. These insights are valuable for planning SSR mitigation and maintaining system stability.

Tech Stack: Python, Classical ML (Logistic Regression, Support Vector machines) and LLM-Hypertuning (Transformers)

Project Description:

This project aimed to distinguish between ChatGPT-generated and human-generated text across multiple domains. Through our research and implementation, we have made significant strides in understanding the unique characteristics of AI-generated content and developing effective detection methods.

• Key findings include:
• Finding key features of text
• Implementation of Classical ML approach (Logistic Regression and SVM) for the binary classification problem
• Implementation of fine-tuning LLM (RoBERTa) for the classification problem

Although we used a standard dataset for train-test splitting, the models achieved high accuracy, with simple Logistic Regression reaching nearly 95%.
The project demonstrates that while fine-tuning RoBERTa for text classification achieves high accuracy on the training data, it struggles with unseen data due to overfitting. Combining RoBERTa’s hidden states with Logistic Regression improves performance on validation data, highlighting the potential of hybrid models.

Tech Stack: Python (Streamlit, Plotly, Altair, NetworkX, Pandas), MATLAB, Data Visualization, Power Systems Analysis, Optimization Algorithms (Exhaustive Search, Particle Swarm Optimization)

Project Description:

• Developed an advanced visualization platform to aid in strategic placement and sizing of Distributed Generation (DG) units within power distribution networks, utilizing IEEE standard bus systems for precise scenario analyses.
• Leveraged Python’s Streamlit interactive tool and data visualization libraries to create an intuitive user interface for visualizing the distribution network, power flow metrics, and the impact of DG integration.
• Implemented robust computational techniques, including Exhaustive Search and Particle Swarm Optimization algorithms, to determine the optimal configuration for DG units, minimizing power loss and ensuring system stability.
• Integrated MATLAB scripts for accurate power flow calculations, taking into account the DG size at each bus, ensuring reliable simulation results.
• Enabled users to adjust DG sizes and locations through a user-friendly interface, dynamically updating the displays and metrics in real-time, facilitating informed decision-making.
• Visualized the relationship between DG size, bus number, and active power loss using interactive 3D mesh grids, highlighting the selected parameters and color-coding metric changes for easy interpretation.
• Animated the optimization process using scatter plots, allowing users to understand the comparative efficiency of their solutions against potential alternatives.
• Provided actionable insights based on calculated optimization strategies, displaying optimal metrics such as bus number, DG size, and revised power losses.
• This project demonstrates proficiency in data visualization, power systems analysis, and optimization techniques, showcasing the ability to develop innovative solutions for efficient and sustainable power distribution networks.

• Implemented the SCADA measurement placement on the PET software for observability analysis.
• MATLAB code developed to facilitate planning and calculation processes utilising Integer Linear Programming. It incorporates observability analysis, sensitivity matrix computation for critical analysis, and cost estimation for installation.

• Found the optimal Distributed Generators (DG) sizes and locations to maintain power loss, voltage stability index, and voltage deviation using analytical and heuristic methods (Optimization algorithms such as PSO, GA etc.) using MATLAB.
• Implemented “Jaya algorithm”, a parameter independent algorithm to solve the optimal DG placement problem which resulted in 75% reduction of power loss.

• Implemented Direct Torque Control (DTC) technique using switching by Space Vector Modulation (SVM) for voltage source inverter (VSI) for an induction motor on Simulink and MATLAB.
• Modeled induction motor for transient and steady-state performance, implemented model DTC scheme to the motor and presented the simulation results.