Solutions

We Live and Breathe Data

Solutions To Fit Your Needs

Whether you are looking for a large-scale private, on-premises data cluster or a storage in the public cloud, we have you covered. Our expertise lies in building large distributed systems and processing real-time data under strict regulatory and security scrutiny. Check out some of our work.

CCAR/FRTB REGULATORY PLATFORMS

INDUSTRY: BANKING

Our custom built CCAR regulatory compliance data lake with dual audit trails underpin the reporting of CCAR to the Fed for two major Wall St Banks.

READ MORE…

REAL-TIME TWITTER EVENT ANALYSIS

INDUSTRY: TRADING

Learn more about how we built a Twitter sentiment analysis engine to capture real- time events for volatility signal generation for HFT algorithms.

READ MORE…

REAL-TIME RECOMMENDATION ENGINE

INDUSTRY: ONLINE GAMING

How do you show that perfect in-game ad to a player in real time? Well, of course you build a recommendation engine based on user profiles.

READ MORE…

AI ENABLED DOCUMENT MANAGEMENT

INDUSTRY: LEGAL TECH

Are you looking for a document manager that is super secure, hosted locally and can automatically extract features and detect anomalies in signed contracts?

READ MORE…

UNSTRUCTURED DATA EXTRACTION

INDUSTRY: OIL & GAS

Do you have a need to extract unstructured data from tons of PDFs like our customer in the shale gas industry? Find out how we used computer vision to help them.

READ MORE…

MARKETING ROI ANALYSIS

INDUSTRY: FMCG

Do you want to figure out how much money you should be spending across your myriad online & offline campaigns, along with advanced visualization of your data?

READ MORE…
CASE STUDY – REGULATORY COMPLIANCE

FINTECH – CCAR/FRTB

The client has to comply with the Federal Reserve’s CCAR requirement. Their global risk platform was built under tight regulatory scrutiny. We had to take millions of records to ensure the timely submission of reports, with processing time dropping from hours to minutes.

XION.AI developed data representations in multiple formats for application optimization. We also provided an interactive end-user analytic modeling interface.

100M+ records processed in seconds
30 billion records from across 30 countries
1500 columns per record
Biggest big data platform at the bank
CASE STUDY – VOLATILITY ENGINE

FINTECH – Volatility Trading Signal Generation

Our client is a large hedge fund company that trades in options on metals and equities. The idea was to look at real-time data feeds coming in from Bloomberg and Twitter feeds, and figure out the sentiment analysis of these feeds.

Based upon that data, we fed the sentiments into the volatility modeling engine that is learning all the time. This is to allow the client to make real time quick decisions at their trading desk with conviction.

Implied Volatility Engine
Risk Modeling Engine
Sentiment Analysis
Real time trading decisions
CASE STUDY – REAL TIME RECOMMENDATION ENGINE

ONLINE GAMING – Real Time Advertising

AI has become essential to providing a more personalized customer experience. Our client was keen to elevate their in-game advertising strategy to understand how and where to insert the right ad into a game scene to drive higher engagement and ROI from the gamer.


We were able to solve this by building a real time recommendation engine that matches the gamer with the right ad. By using data from personal profiles and other credit profile sources, we were able to determine a behavioral profile of the user and segment different users into categories of interests. With that classification in hand, we were able to build a decision tree to predict click through rates for various ads coming through different ad exchanges (doubleclick, admob, FB etc). The recommendation engine thus created had a feedback loop based on predicted vs achieved interaction rates, which was then used to refine the model in real time on our scalable Spark based architecture.

The recommendation engine increased ad interaction rates by over 30% which played a significant role in improving our clients’ bottomline. The project paid for itself in under 2 months.

Cohort Building
Classification & Prediction
Higher Click ratios & Conversion rates
Effective Ad Placement Metrics
CASE STUDY – AI ENABLED DOCUMENT MANAGEMENT

LEGAL TECH – AI Legal Assist

Paralegals spend a lot of time detecting errors and anomalies in contracts, classifying a signed contract before archival, or finding key tables and terms from contracts. Furthermore, lawyers frequently need to assess risk across their contract pool and visualize various elements of their contract database.

With advances in machine learning, we have helped to greatly simplify these tasks with deep learning models that have been trained extensively to automate these tasks and output the data with advanced visualization. The AI Legal Assistant has been created to run completely on premises.

Intuitive & Easy to use interface
Anomaly Detection
Ability to process large volume of contracts
Optimise legal services
CASE STUDY – UNSTRUCTURED DATA EXTRACTION

OIL & GAS – Data Extraction

An oil & gas company uses seismic surveys to produce detailed images of various rock types and their location beneath the Earth’s surface. They utilize this information to determine the location and size of the oil and gas reservoirs before making a decision to purchase the property.

Sound waves are bounced off underground rock formations and the waves that reflect back to the surface are captured by recording sensors, which are then produced in PDF file format.

We were tasked extracting this information from the semi-structured oil well exploration documents. Our platform analyzed more than 100,000 PDF documents, to train standard machine learning models like DF-IDF and our propriety models.

Model adjustments were made on the fly with incoming documents via a feedback loop. The information storage and visualization made it a very practical system in practice.

Fields extracted with dispersed positions in document
Rapidly build on our framework
Easy to use user-interface
Fully traceability of documents to information extraction
CASE STUDY – CLASSIFICATION & PREDICTION

FMCG – Marketing Spend ROI Prediction

A cheese company in upstate New York, the owners of various cheese brands, needed to figure out where to spend their marketing money, be it in online or offline channels.

We looked at their historical data as well as external proprietary data from Nielsen Research and Twitter feeds capturing customer feedback. We built out a customer sentiments analysis engine and fed all the data into a regression model to analyze the marketing spend.

Based on the results, we built a visualization on that allowed the client to view the ROI predictions for the different product range, advertising channels, as well as marketing spend for each of the elements.

Sentiment Analysis
Spend Analysis
Classification & Prediction
Visualization dashboard