Dileep Kumar Sahu

InfraRAG Insight:

• A Retrieval-Augmented Generation (RAG) system leveraging sentence-transformer (mpnet-base-v2) embeddings to extract and summarize infrastructure-related insights from wiki pages, enhancing decision-making with precise, context information in real time.

SMS Text Classifier and Parser:

• Objective was to create customer’s financial profile and leveraging the profile to different teams like personal loan and facilitate bill reminders
• Optimized data models by implementing machine learning algorithms and advanced statistical techniques.

SMS Parser OLAP:

• Developed and implemented SMS-parser OLAP solutions, facilitating real-time aggregation and analysis of resume data.
• Designed and deployed data models for efficient extraction and transformation of SMS resume data into actionable insights.
• Collaborated with cross-functional teams to integrate SMS-parser OLAP solutions, improving data accessibility and decision-making processes.

SMS Parser Dashboard:

• Created customized Looker dashboards to visualize key performance indicators (KPIs) and trends, facilitating data-driven decision-making across departments.
• Implemented advanced data modeling techniques in Lookup to optimize dashboard performance and ensure data accuracy.
• Collaborated with stakeholders to gather requirements and translate business needs into intuitive Looker dashboard designs, improving user adoption and satisfaction.
• Provided training and support to users on navigating and interpreting Looker dashboards, empowering teams to leverage insights for strategic planning and operations.

Device Failure Prediction Model:

• Given historical performance of the device, build an automated model which forecasts the device failure helping quick issue detection and resolution.

Data Room Activity:

• Aim of this project to extract the data (eg. credit score, utility bill, quarterly transaction count, quarterly transactional volume, etc.) for Bajaj finance on common users base.
• Optimized spark configuration to process huge volume of historical data on EMR (Elastic MapReduce).
• Wrote the optimal pyspark script to reduce the execution time as well as reduce the Elastic MapReduce cost(AWS).

PII-Eye:

• PII Model detect the Personal Identifiable Information (PII) from the structured data i.e. data present in the tabular format.
• Identify the personal information from the data, and if PII probability is greater than the threshold value T, then declare that column as PII column.
• Optimized character-level convocational networks (Convects) for text classification.
• Character-level convocational networks could achieve state-of-the-art or competitive results.

User Recommendation System:

• Recommendation Engine was developed and deployed for the Client in Amazon Web Services (AWS), Suggest the similar user interest.
• Hybrid User-based collaborative filtering was developed which helped in identifying the best suited users.
• Recommend the user based on the mutual user implemented using the Network theory.
• Accurate Recommendations so perform well in A/B testing and increase 60% Engagement rate of users.
• Designed, implemented and evaluated new models and rapid software prototypes to solve problems in machine learning and systems engineering.