Mayur Kodche

• Multi-Modal Document Intelligence Platform:
◦ Built end-to-end RAG pipeline using Claude 4.5 Sonnet, LLaVA, and FAISS — processing 10, 000+ legal/healthcare
documents daily at scale:
Achieved 95% accuracy in contract term extraction using hybrid search (semantic + keyword + regex)
Reduced manual review effort by 70% — automating extraction of 70+ fields per contract
Improved table extraction accuracy by 60% using LLM-based OCR correction
Reduced extraction latency by 40% using multithreading and optimized retrieval
Built ingestion pipeline with Unstructured.io + PaddleOCR — handling complex PDF, images, and tables
Deployed on Databricks, migrating to Azure containerized APIs for scalable inference
Key Skills: RAG, Claude 4.5 Sonnet, LLaVA, FAISS, Unstructured io, Padd|eOCR, Databricks, Azure, Python
• Data Pipeline Automation and Monitoring:
◦ Automated monitoring and failure handling for 3, 000+ Airflow DAGS and 1, 000+ Databricks jobs, reducing manual
intervention by 809% and ensuring timely job execution 24/7.
◦ Integrated ServiceNow for automated incident creation, cutting response time by 40% and saving 25, 000 IT hours
annually.
◦ Developed auto-healing mechanisms, improving pipeline stability and reducing downtime by 30%, leading to a 15%
increase in overall system reliability.
Key Skills: Python, SQL(Hive), Apache Airflow, Databricks Workflows, APIs, ServiceNow, Automation
• Chatbot Using RAG Framework:
◦ Developed a chatbot using Mosaic Al Agent Framework, integrated with Slack, and reduced manual query resolution
time by 50%
◦ Leveraged Databricks endpoints and MLflow for model serving, achieving 95% accuracy in real-time user interactions.
◦ Processed 10, 000 unstructured documents using TextSplitter and vector search, improving query response time by
40%.
◦ Automated query handling, reducing manual support efforts by 60% and improving operational efficiency.
Key
•
◦ Integrated Nikes custom Spark Expectations framework with Apache Airflow, reducing data quality issues by 40%
through real-time validation.
◦ Automated error detection and reporting, decreasing
manual intervention by 60% and improving dataset reliability.
•
◦ Automated data validation between SAP and Hive systems, reducing discrepancies
by 50% and improving
accuracy.
◦ Streamlined variance reporting for discrepancies above 30%, cutting resolution time by 35% and enhancing data
governance.
◦ Processed 1M+ records daily using and enabling real-time insights via Tableau dashboards.
◦ a lineage by 30%
◦ Enabled users to view source and target tables for enhancing traceability and efficiency
by 25%

• Architected and deployed an end-to-end RAG pipeline using Claude 4.5 Sonnet, LLaVA, and FAISS — processing
10, 000+ complex legal healthcare documents to date
Developed containerized inference APIs on Azure to host the document intelligence models fo real-time production
serving
Achieved 95% accuracy in contract term extraction using hybrid search (semantic + keyword + regex)
Reduced manual review effort by 70% — automating extraction of 70+ fields per complex contract
• Improved table extraction accuracy by 60% using LLM-based OCR validation and error correction
Reduced extraction latency by 40% using multithreaded parallel execution and optimized vector retrieval
Built robust ingestion pipelines with Unstructured. io + PaddleOCR — handling complex nested PDFs, images, and
tables on Databricks
Key Skills: RAG, Claude 4.5 Sonnet, LLaVA, FAISS, Unstructured. io, PaddIeOCR, Databricks, Azure APIS, Docker, Python
• Data Pipeline Automation and Monitoring:
• Automated monitoring and failure handling for 3, 000+ Airflow DAGS and 1, 000+ Databricks jobs, reducing manual
intervention by 809% and ensuring timely job execution 24/7.
• Integrated ServiceNow for automated incident creation, cutting response time by 40% and saving 25, 000 IT hours
annually.
• Developed auto-healing mechanisms, improving pipeline stability and reducing downtime by 30%, leading to a 15%
increase in overall system reliability.
Key Skills

Data Pipeline Automation and Monitoring =

Automated monitoring and failure handling for 3, 000+ Airflow DAGs and 1, 000+ Databricks jobs using APIs for real-time data retrieval and issue detection.
Enhanced incident management by integrating ServiceNow for automated creation and notifications.
Developed auto-healing mechanisms and optimized workflows to improve pipeline stability and minimize downtime.
Streamlined health checks and alerts to accelerate resolution of critical job failures, saving 25, 000 IT hours annually.
Key Skills: Python, SQL (Hive), Apache Airflow, Databricks Workflows, APIs, ServiceNow, Automation.

DAG/Data Lineage Tracking Tool =

Developed a UI for tracking data lineage between pipelines, improving data transparency and governance.
Enabled users to easily view source and target tables for specific pipelines, enhancing data traceability.
Key Skills: Python, React.js, SQL.

Chatbot Using RAG Framework =

Developed a chatbot using Mosaic AI Agent Framework and integrated with Slack for real-time user interaction.
Leveraged Databricks endpoints and MLflow for model serving and management.
Processed unstructured data with TextSplitter and indexed using vector search for efficient querying and similarity checks.
Automated query handling, reducing manual support and improving operational efficiency.
Key Skills: Python, LLM, RAG Framework, Databricks, Slack Integration, MLflow, TextSplitter, Vector Search.

Spark Expectations for Data Quality =

Integrated Nikes custom Spark Expectations framework with Apache Airflow for real-time data validation, ensuring high-quality datasets.
Automated error detection and reporting, proactively addressing data quality issues and enhancing data reliability.
Key Skills: Python, Apache Spark, Airflow, Nike Spark Expectations, Data Validation, Data Quality.

Data Drift Framework =

Automated data validation between SAP and Hive systems, ensuring timely detection and resolution of discrepancies.
Streamlined variance reporting and ticket generation for variances above 30%, improving data governance.
Utilized PySpark, Databricks, and Tableau for large-scale data processing and interactive dashboards.
Key Skills: Python, SQL, PySpark, Databricks, Tableau, SAP, Hive.

- Data Pipeline Automation and Monitoring:
* Automated monitoring and failure handling for 3, 000+ Airflow DAGs and 1, 000+ Databricks jobs, reducing manual
intervention by 80% and ensuring timely job execution 24/7.
* Integrated ServiceNow for automated incident creation, cutting response time by 40% and saving 25, 000 IT hours
annually.
* Developed auto-healing mechanisms, improving pipeline stability and reducing downtime by 30%, leading to a 15%
increase in overall system reliability.
Key Skills: Python, SQL(Hive), Apache Airflow, Databricks Workflows, APIs, ServiceNow, Automation
- Chatbot Using RAG Framework:
* Developed a chatbot using Mosaic AI Agent Framework, integrated with Slack, and reduced manual query resolution
time by 50% .
* Leveraged Databricks endpoints and MLflow for model serving, achieving 95% accuracy in real-time user interactions.
* Processed 10, 000+ unstructured documents using TextSplitter and vector search, improving query response time by
40% .
* Automated query handling, reducing manual support efforts by 60% and improving operational efficiency.
Key Skills: Python, LLM, RAG Framework, Databricks, Slack Integration, MLflow, TextSplitter, Vector Search
- Spark Expectations for Data Quality:
* Integrated Nike’s custom Spark Expectations framework with Apache Airflow, reducing data quality issues by 40%
through real-time validation.
* Automated error detection and reporting, decreasing manual intervention by 60% and improving dataset reliability.
Key Skills: Python, Apache Spark, Airflow, Nike Spark Expectations, Data Validation, Data Quality
- Data Drift Framework:
* Automated validation between SAP and Hive for 1M+ records daily via PySpark, cutting data discrepancies by 50%
* Streamlined variance reporting for discrepancies over 30%, accelerating operational resolution times by 35%
Key Skills: Python, SQL, PySpark, Databricks, Tableau, SAP, Hive
- DAG/Data Lineage Tracking Tool:
* Developed a UI for tracking data lineage across 500+ pipelines, improving data transparency and reducing debugging
time by 30% .
* Enabled users to view source and target tables for pipelines, enhancing traceability and governance efficiency by 25%.
Key Skills: Python, React.js, SQL

• Gained hands-on experience with Hadoop, Scala, Hive, Apache Spark, and Scoop, while strengthening skills in data
processing, analytics, and big data infrastructure.

• Gained hands-on experience with Hadoop, Scala, Hive, Apache Spark, and Scoop, while strengthening skills in data
processing, analytics, and big data infrastructure.
Key Skills: Hive, Python, Data Processing, Data
• Developed and maintained web applications using SQL, PHP, and front-end technologies, enhancing the learning.
experience with quiz features and chapter videos.
• Built a trainer dashboard for video uploads, quiz management, and attendance tracking, optimizing content delivery.
• Designed and integrated an admin control panel for managing website content and user