Adarsh GorremuchuAI/ML Engineer.

Building autonomous agent workflows and production RAG systems for enterprise Microsoft ecosystems.

• ▸Architected autonomous agent workflows using LLMs, tool-use, and decision logic — cutting manual effort by 70-85% across internal operations.
• ▸Engineered production RAG systems with embeddings and vector databases (OpenSearch, Pinecone), boosting retrieval relevance by 40%.
• ▸Embedded agents into Microsoft ecosystems via Copilot Studio, Semantic Kernel, Azure AI, and Azure OpenAI.
• ▸Devised multi-agent orchestration for classification, planning, retrieval, and structured outputs — accelerated throughput by 45%.
• ▸Strengthened reliability with evaluation checks, access controls, and trace logging — lowered rework by 30% in compliance flows.
• ▸Deployed serverless components on AWS Lambda, SQS, and Step Functions; packaged with Docker; provisioned via Terraform.

Built LLM-assisted analytics and multi-cloud data architecture supporting 20+ research projects.

• ▸Developed LLM-assisted analytics for summarization, classification, and context extraction using RAG and prompt strategies — shortened analysis turnaround by 35%.
• ▸Designed multi-cloud architecture with Snowflake, BigQuery, and NoSQL supporting 20+ projects with standardized schemas.
• ▸Trained and evaluated ML pipelines with TensorFlow, PyTorch, clustering, and statistical methods — elevated forecasting quality by 35%.
• ▸Streamlined orchestration using Airflow, Spark, and Databricks — accelerated pipeline runtimes by 55% while sustaining 99.9% reliability.

Designed ETL pipelines, predictive models, and executive BI dashboards for enterprise reporting.

• ▸Designed ETL pipelines and analytics data models using Databricks, Spark, and warehouses (Redshift, Snowflake, BigQuery) — decreased reporting latency by 55%.
• ▸Applied predictive modeling and KPI forecasting using Python and TensorFlow — increased planning accuracy by 20%.
• ▸Produced executive dashboards in Power BI and standardized KPI definitions — increased stakeholder adoption by 30%.