Ali Dia

Launched BayutGPT, the first large-scale AI product in UAE real estate using Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG). Integrated GPT-based assistants to help users interactively discover properties on Bayut.com

• Increased platform traffic by ~5%, based on post-launch web analytics.
• Improved user engagement and lead generation by ~10%
• Increased net new users by ~7% on the platform

VoiceAgent – AI-driven voice automation platform for vehicle inspections

• Automated phone call scheduling for car inspection appointments using Voice AI Agents
• Reduced manual effort by 70%, freeing up operational teams.
• Increased inspection bookings by ~15% within the first 2 months.
• Designed architecture with Twilio Voice + LLM/NLU model pipeline + webhook automation.
• Implemented fallback logic and escalation systems to handle non-responsive leads.

Leading a cross-functional team of 4 data scientists focused on delivering AI products across real estate and automotive verticals.

• Defined roadmap, assigned tasks, managed sprint delivery
  Facilitated collaboration between product, marketing, and Business intelligence department.

• Execute Listing Text Analysis project using Natural Language Processing (NLP) techniques, including TF-IDF, KeyBERT, and YAKE, to analyze and extract essential keywords (bi-grams and tri-grams) from listing title/description and user chats on the platform and generate custom badges for listings.
• Doubled Adoption Rate: increasing badge adoption from 3% to 6%.
• Generated up to $6,000 in Additional Revenue.
• Boosted Lead Generation listings with new badges vs old ones.
• Improve and maintain ML architecture using a serverless framework on AWS, executing through GitHub actions.
• Establishing Lean ML Standards and implementing CI/CD for the ML Team to enhance system efficiency, reliability, deployment time, and costs.
• Mentoring Intern data scientists and fostering a collaborative team environment.

• Creating feature recommendation engine (content) user/item and item/item on multiple verticals «Properties for Rent/Sale and Classifieds »Based on the similarity of listings and KNN and collaborative recommendation engines for Jobs verticals using ALS.
• 3% Increase in User clicks on the recommended listings displayed on the LPV page.
• Increase in User session time spent on the LPV page.
• Generating a higher Number of leads for recommended listings.
• Developing and validating Machine learning model for the Used cars vertical on dubizzle.com website to ease the appraisal of used cars for dubizzle purchasers and C2C users.
• 7% Uplift in the Profit per transaction with 350k AED Total Net profit increase after deployment.
• User Drop-off rate decreased by sending email campaigns to the users with their car valuation from the AI model.
• Loss transactions on Retail/Auction reduced from 25% to 10%.
• Cars purchasing rate increased from 50 to 150 Cars/Month.
• Increase in Conversion rate of users.

• Implementing RFM analysis and utilizing Beta Geometric Negative Binomial Distribution for CLTV Prediction, and clustering algorithm to categorize users effectively and implemented retargeting strategies based on user clusters.
• Resulting in a 40% increase in user retention.
• Generated Additional Revenue of $2,000.
• Increase in the Net New Leads (NNL) on the platform.
• Enhancing dubizzle.com website by improving category classification for Posting an Add (PAA) using CNN to predict the category leaf based on user Input description.
• Growth in the Conversion Rate of users by 15%.
• 9% Increase in Net New listing on the website.
• Decrease in the Number of Drop-offs ~8% and refunds ~4%.
• Applying an Object detection system on some monetized verticals using a pre-trained model and design/create the architecture with its pipeline (API endpoints).
• Usage of the tool by the Moderation team to identify any presence of multiple objects in one image and moderate it accordingly.
• Prevent Buyers/Sellers on monetized categories from selling different products in one image.
• Deploying ML models under Prod/Dev environment using GitLab and AWS S3 and create dashboards using Sisense Periscope.
• Use of AWS API Gateway to generate endpoints for the tech team.
• Automating ML model pipelines for retraining and performance evaluation.
• Transforming raw data using Redshift and ETL to prepare the data needed.
• Designing lean proofs of concepts (POC) to target business requirements.
• Engaging with stakeholders and proposing solutions based on OKRs.

• Provide real-time business analytics and visualizations for internal customers.
• Build a chatbot using NLP to automate answering new customers' queries on the website.
• Implement and validate predictive modeling algorithm using XGBoost for sales forecasting and matching store inventory with actual needs to reduce storage space and increase profit.

• Support IT Projects with end-to-end system solutions.
• Support VMware deployments and Administration of Microsoft Azure migration on workload.
• Administration of Hyper-V environments for servers.
• Manage servers, monitor health checks, Assess, and implement Disaster Recovery(DR) plans and maintains its life cycle.
• Assess back-ups, GAP analysis, perform tests to make sure data can be retrieved.