VaradKanade

Designed and built an analytics pipeline in Python integrating the RepairShopr API, enabling real-time cross-location performance reporting across 4 store locations.

• Developed a live KPI dashboard and automated daily performance reporting to Slack using Python scripting, eliminating manual reporting work for store managers.
• Built data transformation scripts to calculate employee margins, sales targets, and performance metrics, improving visibility into store-level profitability.
• Combined ongoing software development work with hands-on hardware diagnostics and repair across phones, laptops, and mainboards, maintaining consistent technical delivery across both.

• Developed and tuned ML classification and prediction models in Python (PyTorch, TensorFlow, Scikit-learn, XGBoost) on large-scale industrial IoT sensor data, contributing to up to 45% reduction in plant power consumption.
• Built an equipment degradation-scoring service using Random Forest regression, modelling expected asset behaviour and converting predicted-vs-actual residuals into a continuous, normalized health index, deployed as a FastAPI microservice.
• Implemented a neural survival model (Multi-Task Logistic Regression) in PyTorch to predict Remaining Useful Life from sensor telemetry, producing per-asset survival and hazard curves benchmarked with Concordance Index and Integrated Brier Score.
• Designed large-scale data ingestion, cleaning, and feature-engineering pipelines, applying a custom EDA+ workflow that improved downstream model accuracy by 30%.
• Monitored production model inference with MLflow, tracking performance over time and surfacing degradation signals to trigger retraining cycles.

• Developed multithreaded C++ backend modules using concurrency primitives and optimized data structures, improving runtime performance and system reliability.
• Debugged production issues including memory defects, race conditions, and concurrent execution bottlenecks, improving stability of backend and system-level components.
• Optimized SQL queries across PostgreSQL and SQL Server through query plan analysis, reducing execution time by 25% on large datasets.
• Implemented JWT-based authentication and secure access handling for business-critical backend APIs.
• Participated in code reviews and root-cause analysis within an Agile team, contributing to maintainable, production-ready backend components.