Our Impactful
Projects
From smart automation to advanced analytics, our projects showcase how we apply data science, AI/ML, and data engineering to solve real-world problems and drive meaningful outcomes across industries.
Client Persona
Japanese netsourcing company specialising in designing, integrating, and managing network infrastructure and traffic monitoring systems.
Challenges
The client needed a fast and reliable way to retrieve accurate SOPs from historical case data written in Japanese to ensure consistent and efficient issue resolution.
Solution
- Developed machine learning models to generate SOPs based on input case data.
- Enabled near-instant retrieval of relevant SOPs, reducing search time from hours to seconds.
Tech Stack
- Python
- PyTorch
- LangChain
- Weaviate
- GCP
- Huggingface
AI-Powered Dashboard for Insurance Insights
Client Persona
U.S.-based insurance company integrating AI-driven dashboards and intelligent money routing systems.
Challenges
Business users needed a fast, intuitive way to generate visual insights from complex payment and insurance data—without relying on technical teams or manual reporting.
Solution
- Developed an AI-powered interactive dashboard using natural language prompts.
- Enabled instant generation of charts and insights from insurance and payment data.
- Empowered users to explore data independently, accelerating decision-making.
Tech Stack
- Python
- LangChain
- GPT-4.0
- Gemini 3.5
- Claude
- PostgreSQL
- Azure
- Agentic Pipeline
- FastAPI
Entity Detection for Trade Intelligence
Client Persona
Canadian trade intelligence provider offering cloud-based logistics and supply chain solutions.
Challenges
The client needed a reliable method to identify organisation names from unfiltered text data to streamline restricted party screening and compliance workflows.
Solution
- Developed an AI-powered classifier using advanced NLP techniques.
- Trained to detect organisation names within raw text.
- Achieved an 80% reduction in scanning time, significantly improving operational efficiency.
Tech Stack
- Python
- PyTorch
- Scikit-learn
- Azure
Retail Forecasting ETL Framework
Client Persona
U.S.-based product company building forecasting systems for supply chain clients.
Challenges
- ETL pipeline took over 5 hours to process billions of records.
- Rigid transformation logic slowed adaptability and analytics.
Solution
- Rebuilt ETL using PySpark on Apache Spark.
- Introduced a config-driven transformation framework.
- Cut processing time to 45 minutes and improved scalability.
Tech Stack
- Python
- PySpark
- Apache Spark
- Azure
License Screening & Product Classification for Trade Intelligence
Client Persona
Canadian trade intelligence provider offering cloud-based logistics and supply chain solutions.
Challenges
- Frequent changes in U.S. export regulations due to geopolitical shifts.
- Needed a reliable way to determine licensing eligibility for international shipments.
- Required accurate product classification from thousands of entries based on text descriptions.
Solution
- Built a rules-based engine atop custom NLP models to assess licensing requirements.
- Developed intelligent search models to classify products from large libraries using textual input.
- Enabled faster, regulation-compliant decisions for global trade operations.
Tech Stack
- Python
- TensorFlow
- SpaCy
- Scikit-learn
- Oracle Cloud
False Positive Reduction in Restricted Party Screening
Client Persona
Canadian trade intelligence provider offering cloud-based logistics and supply chain solutions.
Challenges
- High-volume customer data with inconsistent quality led to excessive false positives in restricted party screening.
- Wide search parameters ensured true matches but overwhelmed users with low-quality results.
Solution
- Built a rules-based engine using Python to intelligently filter out low-quality matches.
- Reduced screening volume from thousands to a few hundred names, improving efficiency and user experience.
Tech Stack
- Python
- FastAPI
- PolyFuzz
- TFIDFVectorizer
- Azure
Collision Prediction Using Dashcam Footage
Client Persona
Europe-based dashcam company focused on enhancing road safety through intelligent video analytics.
Challenges
Develop a proof of concept (POC) to predict vehicle collisions in real time using dashcam footage, accounting for:
- Diverse weather conditions
- Occlusions
- Unexpected road events
Solution
- Built a CNN model that analyses video input to predict collision probability.
- Achieved 80% accuracy in identifying potential collisions before they occur.
Tech Stack
- Python
- PyTorch
- YoloV8
- 3D_Resnet
Real-Time Mining Dashboard Implementation
Client Persona
U.S.-based wireless device company designing integrated systems to enhance safety, efficiency, and productivity in mining operations.
Challenges
- No unified system to monitor underground miner activity and production metrics in real time.
- Limited visibility led to safety risks, inefficiency, and delayed decision-making.
Solution
- Built a scalable data engineering pipeline using PySpark on Apache Spark.
- Ingested and processed real-time data from miner tracking and production systems.
- Delivered interactive dashboards with live insights into operations, boosting safety and transparency.
Tech Stack
- Python
- PySpark
- Apache Spark
- Azure
Snowflake Data Migration
Client Persona
A U.S.-based healthcare organization serving 500,000+ individuals with free medical, dental, mental health, and prescription services.
Challenges
Legacy SQL Server infrastructure limited scalability, performance, and integration with modern analytics tools.
Solution
- Built a secure ETL pipeline using Azure Data Pipelines and Python
- Migrated data to Snowflake for scalable, cloud-native access
- Enabled faster, more reliable insights across the organization
Tech Stack
- Python
- Azure Pipelines
- Azure Functions
- Snowflake
- Azure Cloud
Data Collection Platform for Academic Institutions
Client Persona
India based education provider offering:
- Course & Student Management
- Content & Test Delivery
- Live Lectures
Challenges
Disorganized data on institutions across sectors made it hard to guide students accurately.
Solution
Built a Python pipeline to:
- Scrape and clean data (Beautiful Soup, Selenium)
- Store structured info in SQL Server
- Enable personalized admission guidance
Tech Stack
- Python
- Beautiful Soup
- Selenium
- SQL Server