Build intelligent applications that drive business success with AI. 🚀

Achievements

• Brevo — Built a fully automated multi-agent customer support system with the CEO & CX team: multimodal RAG, voice-call transcription, and sentiment analysis.
• Ragmaker — Developed an end-to-end agentic platform unifying queries across multi-modal documents and structured databases, leveraging MCP and modular agent skills for business orchestration at scale.
• CityPark — Built an AI-powered reservation management system automating booking email processing from 10+ parking platforms (Free2Move, Parkos, Parkivé) with custom NLP algorithms and LLM extractors.
• Beyrey Ads — Designed an AI-driven AdTech solution on Azure integrating GenAI for automated marketing content creation (Instagram, LinkedIn) and a text-to-SQL Maker Agent for campaign insights.

Skills Used

• Python · FastAPI · FastMCP
• Langchain · Langgraph
• RAG · Multi-Agent
• NLP · LLM
• SQL · Docker

Tools

• Azure (AI Search, Functions)
• OpenAI API
• Docker · Render
• Git · GitHub

At Mantu Group, I disseminated knowledge on LLM technologies, influencing adoption and application strategies across business sectors.

Achievements

• Directed the development of a multi-agent LLM application, enhancing data extraction, summarization, and tagging capabilities.
• Managed Azure-based infrastructures, optimizing resource utilization and performance for critical data operations.
• Implemented a RAG system on Azure, improving the F1-score of a GenAI summarization agent from 80% to 96%.
• Engineered a Python bot for systematic data collection, supporting advanced AI analytics and decision-making frameworks.
• Enhanced machine learning operations by integrating a Generalized Linear Model, achieving 98% explained variance.
• Pioneered an XGBoost model to estimate CO2 emissions, achieving a 15% error rate, aligning with European environmental standards.
• Developed a GenAI paradigm for telecom, automating customer segmentation across 120 groups via a dual-agent system.
• Designed a serverless AWS architecture schema for KSL Singapore to integrate GenAI into its application.

Skills Used

• Python · Langchain · Langgraph
• RAG · Azure AI Search
• Vertex AI · SageMaker
• XGBoost · Random Forest · GLM
• Multi-Agent Systems

Tools

• Azure (AI Search, ML Studio)
• GCP (Vertex AI, BigQuery)
• AWS (Lambda, S3, SageMaker)
• Dataiku · Git

• Built regression models to predict customer behavior trends, providing valuable insights for the operational team.
• Implemented an ETL process using Python, significantly improving the speed and accuracy of data processing.
• Performed statistical analysis to create clusters of treatment folders based on duration using hierarchical clustering.
• Created a Power BI dashboard for dynamic analysis, leveraging new KPIs such as recovery factor and Sharpe ratio for trend following.
• Developed a data architecture using SQLite databases and engines.
• Automated data flow between on-premises servers and Power BI dashboards.
• Enhanced data processing speed by 3000% through advanced data architecture and ETL pipelines.
• Designed a BI tool to monitor country-level recovery performance, incorporating new statistical KPIs like Large Loss Factors, Sharpe Ratio, and Lifetime Value.

Skills Used

• Python
• SQL

Tools

• Hadoop
• Big Query
• Visual Studio Code
• Git
• AWS
• Power BI
• SQLite

• Led the mining of external data for the 'Golden Table' project to optimize content acquisition strategies, enhancing the media catalog's value and audience reach.
• Conducted A/B testing to determine animation watching factors on the streaming platform.
• Developed a churn model with 99% accuracy, significantly reducing customer attrition, saving 3-4K users daily.
• Applied common statistical applications (A/B testing, probabilities, regression) to evaluate the performance of the 6play platform.
• Experience with SQL and working with large data sets (Hadoop/AWS/Big Query) to characterize users (profile segmentation) and respond to business requests.
• Utilized machine learning to develop prediction tools (classification and regression models) for predicting user loss volume.
• Applied time series analysis to user consent for analytics and targeting purposes following GDPR reforms.
• Performed root cause analysis (RCA) by testing K-populations around the same KPI extracted from a combination of N variables.
• Onboarded and presented findings to various business teams, linking study results to business KPIs and demonstrating model performance on retention and reactivation rates.

Skills Used

• Python
• SQL

Tools

• Hadoop
• Big Query
• Visual Studio Code
• Git
• AWS