logo logo


Sukhbir Benipal

Solr Consulting

Our Solr Consulting and Implementation Services: ● Translating business requirement into a Solr cluster implementation. ● Creating Solr queries for the best search experience. ● Making it Fast. Scalable. Improving performance and scalability of an existing Solr cluster. ● Deploying on AWS or Custom Hardware for maximum performance and traffic spike adjustability. ● Tuning relevance ranking using various combinations. Slow and Patient work. Our Pricing: Fixed Rate to begin. Hourly thereafter.

About Benipal's previous Work

Shopping Search Engine: 300 Million Products. 1.2B Titles and Descriptions. 1B+ Images. 12,000 Merchants.
● Voice Search and Image matching, recognition and search.
● Highly scalable infrastructure with average response times under 100ms.
● Contextual + Relational, neural network based Shopping Search Engine able to understand user queries and provide exact results for “ Blue Bedspread by Martha Stewart from Walmart or Macy's.com or around me for under $500”.
● High Volume Search + Big Data Infrastructure allowing Products and Search Queries to reflect most recent state.
● Built and Managed 40 High Performance Servers in Datacenter with 10G uplinks.

Search Engine Architect / AI Researcher / Supercomputing ● Added partial NLP to search, letting the computer “understand” the query.
● Created a self-healing, self-learning “Auto Product categorization” algorithm that can automatically analyse and categorize products in any of over 30,000 available categories. Successfully used and demonstrated success rate of around 85%.
● Built a 20 Tflops CPU based Super Computer with over 12TB RAM, 640 Cores and 1 PB Storage. Easy to add GPUs to increase total Floating Point computation.
● Theorized and Worked on Computer Vision with small GPU based cluster to provide a better understanding of how neural networks can understand images and “see” Videos.
● Theorized and Researched on Artificial Intelligence using various current open source projects and how integrated usage could provide a better understanding of neural networks and their application to live real world data scenarios by providing computers with the ability to “understand” different datasets, “see” images and videos and roughly match their interconnects.