Gartner predicts
that in 5 years,
80% of all analytics will be
done on Graphs!
WHY ZETTABOLT
Working with the leading Graph DB provider (Tiger Graph) and hardware acceleration giants (nVidia and AMD) has exposed us to some of the toughest challenges faced by top Finance, Pharma and IT companies in using graph analytics. Having delivered projects which can speedup credit-card fraud detection by over 100x, we understand what it takes to build performance and cost efficient graph based solutions.
OUR GRAPH
SKILLS
Graph use modes : finance, social, pharma, network/IT
Deep knowledge of Tiger Graph, neo4j etc
Migrating SQL DBs to Graph DBs
Feature engineering and model performance tuning
WRITING CUSTOM UDFs/
KERNELS FOR CPU, GPU, FPGA
CPU:
-
Optimizing the compute using SIMD, vector processing and custom algorithms.
-
Simplifying the compute flow to remove redundancies.
GPU/FPGA:
-
Writing hardware kernels based on customer’s requirements.
-
HW/SW partitioning for best utilization of GPUs/FPGAs
Improving FPGA kernels (SQL operations, Graph Algorithms etc.)
GNN
SERVICES
From a basic model created by Data Scientist, we can create a production GNN based model creation setup. Flow can work on multi-node-multi-GPU clusters (on-prem or cloud) and scale to very large training sets.
OUR BLOGS & WHITEPAPERS
