LIRILKUMAR AMAL

Played a pivotal role in building the Personlised.AI product, supporting backend operations for marketing campaigns, deploying AI modules, and training recommendation models for major QSR clients, leading to significant revenue increases.

• Item Recommender: Developed a model to recommend top personalized products to customers.
• Optimized Offer Recommendation: Created a system to suggest profitable offer amounts for marketing campaigns.
• Customer Segmentation: Implemented models to segment customers for targeted marketing campaigns based on their life cycle stages.
• Propensity Models: Built models to predict customer behavior, including churn, order, and offer redemption likelihood.
• Market Basket Analysis: Conducted association rule mining to identify and recommend items frequently purchased together for cross-sale campaigns.
• Look-Alike Modeling: Designed models to identify target audiences for campaigns by finding customers similar to a defined seed group.

Owned Product Deployments at Clients

• Actively contributed to planning various marketing campaigns focused on cross sell, upsell and retention, starting from digital app adoption, customer retention and customer churn mitigation, customer stickiness and new item adoption
• Worked with customer success team to measure campaign impact using various AB testing analytical groups.

Trained junior data scientists, machine learning engineers to scale up and sharpen their skillset.

Developed end to end tailor made solutions for various internal teams including services, sourcing and supply chain.

• Signature Extraction and Identification: Developed a product for locating and extracting signatures from legal documents using classification, object detection (YOLOv5), and signature matching models.
• Demand Forecast Product: Built a supply chain product to forecast hardware demand, optimizing production and reducing inventory costs using XG boost Regressor.
• Code Reviewer Recommendation Tool: Designed and deployed a flask microservice offering hybrid recommender model (LightFM) to suggest reviewers for open code change requests.

• Wrote modular python ETL pipeline feeding data into machine learning models predicting water demand at reservoirs i.e., Exposed models as API using Flask to web application.