Our Projects

KUDO

KUDO is an innovative multilingual web conferencing platform equipped with both human and Al- powered live interpretation capabilities. Seamlessly deliver real-time interpretation to enhance the quality of your conferences. The KUDO interpreter marketplace offers a streamlined booking process, granting you convenient access to a vast network of interpreters available on-demand.

What did we do?

• We used NLP techniques to automate the process and achieve maximum efficiency.

• Developed NLP pipeline for Automatic Terminology Extraction and Named Entity Recognition that used spacy models.

Tech Stack

• Python

• Django

• Flask

• NLP

Punchh

The Punchh platform is an innovative and cutting-edge ML-powered customer engagement toolkit that excels in crafting unparalleled retention and remarketing experiences that seamlessly traverse the entire customer journey, from the very first warm greeting to the hundredth enthusiastic welcome back, ensuring a remarkable and enduring connection between businesses and their valued clientele.

What did we do?

• Built end-to-end data pipelines for handling large amounts of data using Databricks, Snowflake and Kafka, PySpark, and Python for fetching, processing, and streaming data.

Tech Stack

• Python

• Pyspark

• Kafka

• Snowflake

• Databricks

Muse

Muse analyzes real time transaction data so that financial service providers can offer expert financial insights, increase engagement and boost customers' finances. Muse sets the stage for new financial services products by adding real-time analysis of your customers' transactions. Customers will know which transactions provide tax saving opportunities by simply connecting their accounts.

What did we do?

• Developed OCR to read the information off tax returns and utilize a set of rules that shows the most effective ways and steps that users can perform to save taxes.

Tech Stack

• Python

• Django

• Tensorflow

• SLKearn

• AWS

ParseAI

Parse Al brings innovation and speed to real property transactions by leveraging machine learning (ML) and optical character recognition (OCR) to assist with title research. This platform can rapidly extract critical information from a large collection of title documents to allow title researchers to complete abstracts and reports faster and more accurately.

What did we do?

• Developed machine learning and OCR algorithms using Python, TensorFlow, and Keras to aid in title research and extract information from the title documents.

Tech Stack

• Python

• Tensorflow

• Keras

Terrascope

A highly sophisticated property portal harnessing the power of state-of-the-art feedback-based machine learning algorithms, this platform is designed to offer users an unparalleled property search experience by meticulously tailoring property recommendations to their individual search preferences and incorporating real-time feedback to continuously refine and enhance the search results.

What did we do?

• Developed collaborative as well as content-based algorithms for recommending listings to users.

• Created a computer vision service using UNetFormer implemented with Python that performs segmentation on satellite images highlighting features such as vegetation cover, water bodies, pavements, and residential as well as commercial structures.

• The recommender system was developed using Python, Pandas, and sci-kit-learn to score properties and run custom scripts on elastic search. It used content-based as well as collaborative filtering to recommend properties.

Tech Stack

• Python

• UNetFormer

• Pandas

• Sci-Kit-Learn

RedFlag

Redflag develops software that is able to analyze every type of content used to communicate online (text, image, video, and audio) and has the capability to find any particular piece of content across the entire internet. It provides solutions that allow businesses and individuals to both protect their content and fully understand their online reputation so that it can be properly managed. These products are utilized by enterprises and individuals across media, entertainment, sports, publishing, music, retail, finance, and beyond.

What did we do?

• Developed advanced data extraction and preprocessing techniques to gather data from social platforms and apply in-house natural language and computer vision models developed using Python, SKLearn, NLTK, Spacy, and Keras to detect piracy and copyright infringement.

• We also utilized data mining techniques to extract profile insights for users.

Tech Stack

• Python

• SLKearn

• NLKT

• Spacy

• Keras