Prasad Dharaskar

MAQ Software · Hyderabad, India

• Contributed to a LinkedIn contractor engagement, delivering production-grade features across full-stack and data platforms.
• Built an AI-powered support assistant on Azure AI Foundry integrating Classifier and Knowledge Search agents with Azure OpenAI (GPT-4o) and AI Content Safety for grounded, safe responses.
• Designed and built custom MCP (Model Context Protocol) tools/servers callable directly from Azure AI Foundry agents, exposing enterprise actions and data sources as first-class tools.
• Integrated an agent framework to orchestrate Foundry-deployed models within an App Service backend — giving the App Service a "brain" for reasoning, tool calls, and grounded responses.
• Developed Azure Function Apps (custom MCP Server + indexing pipeline) syncing Dataverse tables into Azure AI Search, exposed via an App Service streaming chat endpoint consumed by a Power Platform portal.
• Secured services with Entra ID & Managed Identity; instrumented Application Insights telemetry for latency, conversations, and response quality.
• Operated as SRE owning host health, provisioning, and OS upgrades; migrated MAAS URLs (MAAS2→MAAS3) and built GitHub Actions CI/CD.
• Built a Jupyter Notebook extension to securely query HDFS/Hive via Py4J, packaged and tested with Docker.
• Developed a Python automation tool that cloned 100+ models and applied DBT configs, saving 40 hours/month.

MAQ Software · Hyderabad, India

• Delivered full-stack features using C#, .NET, ASP.NET Core, React.js, and Node.js with clean, maintainable code.
• Designed batch and real-time data pipelines with Hadoop, HDFS, Kafka, Spark, PySpark, Hive, and Airflow DAGs; automated ingestion (Kafka→HDFS→Hive) with Linux shell scripting for ETL.

Indian Institute of Information Technology, Pune · CGPA 7.56