Our Projects
AI Data-driven solutions to ensure optimal results.
Ingesting Telemetry from Smart Buildings into Digital Twins for Live Monitoring and AI predictions
Challenge
When receiving millions of measurements from IoT devices (power consumption, vibration, temperature, people traffic, etc.) from different buildings, how can one provide meaningful live insights and predictions to feed a digital twin?
Value Delivered
Every facilities manager was given a control panel to monitor online all indicators from their respective building. Alerts were created when certain indicators reached predetermined thresholds. Predictive maintenance insights from machine learning models were developed, such as those for elevator vibrations.
Technical Solution
We successfully managed the continuous streaming of millions of telemetry records from multiple buildings, leveraging the robust capabilities of Azure Synapse Analytics pipelines for data ingestion. Subsequently, we executed Extract, Transform, and Load (ETL) operations to appropriately refine and prepare the data. We then used Azure Machine Learning to generate predictions. Finally, we used Power BI Embedded to create sophisticated visualizations, facilitating real-time analysis and insights.
Factory Line Productivity Analysis
Challenge
How can one monitor the productivity of each line and individual in a factory setting?
Value Delivered
A live dashboard tracking productivity was provided to each line manager. Insights on key influencers of productivity declines were shared, fostering continuous improvement on the factory line.
Technical Solution
We connected to the On-Premises SQL Server database, which captures factory measurements using the PBI Gateway, and published reports in Power BI providing live data.
Customer service with ChatGPT
Challenge
Optimize the efficacy and cost-efficiency of the customer service team in scenarios that necessitate an in-depth comprehension of internal products and services.
Value Delivered
The introduction of a ChatGPT bot is proposed to address a substantial 80% of customer service inquiries. This approach would facilitate the customer service team's focus on the remaining, more complex 20% of inquiries, consequently maximizing the value delivery in these higher-level interactions.
Technical Solution
The ChatGPT bot underwent extensive training on internal documentation pertaining to products and services via Azure OpenAI. This process ensured not only a high level of expertise in the bot's responses but also the preservation of data confidentiality.
Process automation with LowCode Apps using ChatGPT
Challenge
Establish automated processes within the Legal Department, an area often marginalized by IT due to conflicting investment priorities.
Value Delivered
Through the implementation of a specialized application, the Legal Department's operational efficacy was significantly optimized, realizing a time-saving efficiency of over 20% in their daily tasks.
Technical Solution
The devised solution entailed the development of a multi-faceted application constructed around three integral components: Power Apps for data input, Power BI for detailed reporting, and Power Automate for proficient data extraction. Further augmentation came with the integration of Power Apps Copilot, powered by ChatGPT, facilitating a LowCode environment. This approach contributed to a substantial reduction in both development expenditure and time to market.
Google Analytics for Food Industry
Challenge
How to ingest Google Analytics 4 data from Food Industry client and create valuable insights
to optimize campaigns.
Value Delivered
Our solution enabled the client to gain a comprehensive understanding of the impact of social networks, leading to significant enhancements in campaign effectiveness.
Technical Solution
We connected to GA4 API from Airbyte for “out of the box” API calls and data ingestion and with Airflow for advanced customized API calls and transformations. Data was then onboarded onto a Snowflake datawarehouse and from there visualized in Power BI.
Evaluating Impact of Social Networks & Salesforce Campaigns
Challenge
In a world where marketing budgets in the millions are invested across various social networks, the question arises - how can one effectively track and compare the return on investment and campaign efficacy across diverse platforms, time periods, and countries?
Value Delivered
Our solution enabled the client to gain a comprehensive understanding of the impact of social networks, leading to significant enhancements in campaign effectiveness. It empowered them to discern where the Return on Ad Spend (ROAS) was higher – i.e., Instagram or TikTok. Moreover, each country could view its own investments and benchmark them against the effectiveness of other countries. This facilitated the sharing of lessons learned across geographies.
Technical Solution
We connected to multiple data sources such as Facebook, Instagram, GoogleAds, Google Analytics, LinkedIn, etc., utilizing their respective APIs, with Azure Synapse notebooks developed in Python. We orchestrated daily data ingestion and performed ETL for data cleaning and preparation, storing it in Azure Synapse Datawarehouse. This DW consolidates data from all global social networks. Power BI visualizations were developed and read from it. Record Level Security was implemented, ensuring users from different countries accessed only their relevant data.
Price Predictor for Complex Products
Challenge
For complex products with numerous options and variations, such as a car, how can one create a predictive model that guides the pricing strategy, taking into account competitors' prices?
Value Delivered
We developed a predictive model to support the brand's pricing strategy, enabling pricing analysts to enhance their targeting and effectiveness.
Technical Solution
We connected to multiple data sources such as Facebook, Instagram, GoogleAds, Google Analytics, LinkedIn, etc., utilizing their respective APIs, with Azure Synapse notebooks developed in Python. We orchestrated daily data ingestion and performed ETL for data cleaning and preparation, storing it in Azure Synapse Datawarehouse. This DW consolidates data from all global social networks. Power BI visualizations were developed and read from it. Record Level Security was implemented, ensuring users from different countries accessed only their relevant data.
Transitioning Financial Reporting from Excel & PowerPoint to Power BI
Challenge
What is the best approach to generate financial reporting that meets management expectations? How can one automate the generation and distribution of the monthly financial deck?
Value Delivered
Management was provided with the desired reporting to support decision-making, yielding significant savings in the financial team's time. We introduced a security layer to ensure each stakeholder could only access data from
their respective country. Additionally, an audit trail was established to track the distribution and data access.
Technical Solution
We connected to the Microsoft Dynamics database using the SQL Server Airbyte connector and
uploaded relevant tables into Snowflake daily. We then created multiple Power BI reports in line with
financial management expectations, all reading from Snowflake.
Monitoring Twitch's Top Streamers
Challenge
Navigating the rapidly changing landscape of Twitch, how can one effectively keep track of the top streamers?
Value Delivered
We developed a dynamic live dashboard that provides real-time tracking of the top streamers, categories, and videos on Twitch. This interactive tool empowers users to stay abreast of the latest trends and movements within the Twitch community.
Technical Solution
We leveraged a Python notebook in Azure Synapse to interface with the Twitch API, ensuring seamless data retrieval. The gathered data was subsequently loaded into the Synapse Data Warehouse. From there, it was channeled into Power BI, fostering an efficient data flow pipeline from Twitch to our live dashboard.