Tấn Đạt Lê

Data Engineer with 3 years of experience building high-performance ETL pipelines and scalable Data Warehouses for Batch and Real-time processing. Skilled in driving cross-functional collaboration to support AI/ML data initiatives, with hands-on expertise in Generative AI integration, Python, dbt, Spark, and Cloud Infrastructure (AWS/GCP).

Data Analytics Engineer at Momo (2024-10 – Present)

• Developed an automated migration tool to transition ETL workloads from BigQuery to a Spark/Hive Data Lakehouse, projected to reduce monthly compute costs by ~40%.
• Co-engineered a SQL Transpiler using Python to automatically convert complex BigQuery SQL syntax into SparkSQL, reducing manual refactoring effort per pipeline from 3 days to under 30 minutes.
• Orchestrated the end-to-end migration pipeline using Temporal, managing the full lifecycle from SQL transpilation to data validation, ensuring atomic execution and zero data discrepancies between systems.
• Integrated Generative AI models to parse legacy DAG source code for metadata extraction and auto-correct transpiled SQL errors, reducing execution failure rates by 90%.
• Collaborated with AI and Data Science teams to facilitate the seamless migration of critical training datasets to the Lakehouse, ensuring zero downtime for ongoing model development workflows.
• Architected a modular backend system using FastAPI and PostgreSQL on Kubernetes (GKE), establishing a scalable platform for automated data collection, extraction, and analysis.
• Implemented AI Agents using LangChain to automate administrative workflows (e.g., parsing raw text, extracting insights, and summarizing logs), improving operational efficiency.
• Engineered a Text-to-SQL chatbot utilizing LangGraph and Vertex AI that empowered business users to self-serve data, reducing ad-hoc SQL requests by 40% and cutting time-to-insight from hours to seconds.
• Established governed prompt standards, model routing, and high-throughput agent execution pipelines, scaling orchestration to 200+ daily Airflow pipelines.
• Designed agent observability and monitoring systems to track AI behavior, cost, and pipeline performance at scale.
• Collaborated with cross-functional teams to define strict Data Contracts, ensuring consistency of input features for downstream AI/ML models.
• Engineered automated data quality workflows using dbt and Airflow, sustaining a 95%+ quality score across 200+ critical financial data tables.
• Wrote custom Python utility scripts to auto-generate complex SQL validation files, eliminating 80% of manual test creation effort and ensuring 100% test coverage.
• Configured GitLab CI/CD pipelines and Docker containers to automate the testing environment, preventing broken code from disrupting production workflows.
• Developed a real-time anomaly detection function using GCP Cloud Functions to catch schema drifts and freshness violations, preventing 98% of downstream AI/ML pipeline failures.
• Deployed and customized the OpenMetadata platform, contributing to a centralized Source of Truth for the Data Lakehouse that increased overall data discoverability by 80%.
• Implemented backend extensions using Python SDKs to support column-level lineage and SQL query parsing, providing granular visibility into data dependencies.
• Optimized metadata ingestion pipelines using Apache Airflow to index over 300,000+ tables and track 15TB of daily processing volume, strictly maintaining T-1 data freshness SLAs.
• Integrated a Generative AI feature using LLMs to auto-generate semantic descriptions for tables and columns, achieving 16% documentation coverage for assets previously lacking metadata.

Data Engineer at Sagan Technologies Group - Outsourcing (2023-08 – 2024-09)

• Contributed to the development of the Customer Data Platform on AWS and Hive, consolidating data sources to build a unified Feature Store for customer analytics.
• Partnered with Data Scientists to build scalable feature engineering pipelines using PySpark, transforming raw logs into structured features for churn prediction and segmentation models.
• Integrated machine learning inference steps into Airflow DAGs using scikit-learn and Pandas, automating the generation of daily customer insights for marketing teams.
• Set up Docker containers and Git CI/CD workflows to automate testing, reducing the time-to-market for new ML features by 40%.
• Built automated data ingestion pipelines using Python and Apache Airflow to stream and process highly complex, multi-page unstructured clinical documents from AWS S3.
• Integrated a Named Entity Recognition (NER) model to extract and classify clinical indicators from raw text, transforming unstructured patient data into structured PostgreSQL records.
• Implemented robust data validation and error handling scripts, which ensured data integrity and reduced downstream analysis errors by 95%.
• Developed high-performance REST APIs using FastAPI to serve health insights to the frontend, containerized the application using Docker and Kubernetes for scalable deployment.
• Collaborated with Data Scientists to design and implement scalable data pipelines on AWS, enabling efficient real-time processing of medical indicators.
• Partnered with Data Scientists to construct scalable data pipelines on GCP using Cloud Composer, Pub/Sub, and BigQuery, ensuring the continuous flow of real-time trip signals.
• Developed distributed data processing jobs using PySpark to clean and standardize historical booking records, handling missing values and duplicates for high-quality model training.
• Assisted in feature engineering and model validation for XGBoost and LGB algorithms, transforming raw time-series data into actionable predictive insights.
• Built interactive dashboards in Tableau to visualize demand heatmaps, empowering operations teams to optimize vehicle allocation efficiency by 15–20%.

Bachelor of Science in Computer Science (AI Specialization) – HCM University of Science (2019-08 – 2023-10)