M
Maha Alton
About
Detail
New City, New York, United States
Résumé
Jobs
verified_user
0% verified
Senior AI/ML EngineerCrestaNov 2024 - Current (1 year 11 months)• Designed and deployed scalable ML pipelines on AWS to process 500M+ records per month, supporting both
structured and unstructured data, reducing training runtime from days to hours.
• Built and optimized PySpark + Apache Spark workflows for large-scale NLP model training, enabling real-time
analysis of 1M+ customer interactions daily.
• Developed LLM-based solutions (RAG pipelines, prompt engineering) to automatically summarize and classify
200k+ documents monthly, increasing knowledge retrieval efficiency.
• Orchestrated end-to-end ML workflows using Apache Airflow, implementing automated scheduling and
alerting that prevented over 150 pipeline failures per month.
• Deployed ML models with FastAPI- AI/ML EngineerINFUSEMay 2021 - Oct 2024 (3 years 6 months)• Developed and implemented machine learning models for classification, regression, and clustering using
Python, Scikit-learn, and XGBoost, handling datasets of up to 50M records.
• Built and maintained data pipelines using PySpark, Apache Spark, and SQL to process terabytes of structured
and unstructured data.
• Worked on NLP use cases using Hugging Face, BERT, and Transformer models for text analytics and sentiment
analysis, processing over 100k customer reviews per month.
• Orchestrated ML workflows using Apache Airflow for automated training, scheduling, and reporting,
preventing 100+ workflow failures per month.
• Deployed models using FastAPI and Docker, supporting batch and near real-time infer - Junior AI EngineerKlaviyoMar 2018 - Apr 2021 (3 years 2 months)• Assisted in building machine learning models for predictive analytics using Python, Scikit-learn, and Pandas.
• Processed and cleaned large datasets using SQL, PySpark, and Python to support model development.
• Implemented basic NLP tasks including text preprocessing, sentiment analysis, and topic modeling using
Hugging Face.
• Supported model evaluation, feature engineering, and hyperparameter tuning to improve performance.
• Assisted in workflow automation using Apache Airflow and supported model deployment with Docker.
Education
verified_user
0% verified
- BBachelor's in Computer Science
Projects (professional or personal)
verified_user
0% verified
- SScalable LLM-powered conversational analytics platformBuilt a scalable LLM-powered conversational analytics platform using RAG architecture (LangChain,
LlamaIndex), integrating vector databases (Pinecone, FAISS) for semantic search across 10M+ enterprise
documents. Implemented document ingestion pipelines using PySpark and Apache Airflow, enabling real-
time knowledge retrieval and context-aware responses via GPT-based models. Optimized embeddings and
retrieval strategies to reduce response latency by 40% and improve answer accuracy. Deployed the solution
using FastAPI, Docker, Kubernetes on AWS (EKS, S3), with monitoring via MLflow and Prometheus for
performance tracking and observability. - RReal-time ML pipelines for fraud detectionBuilt real-time ML pipelines for fraud detection on high-volume transaction streams using Apache Kafka,
Spark Structured Streaming, and PySpark. Developed feature extraction and scoring pipelines to detect
anomalies in near real-time, deployed classification models using TensorFlow, PyTorch, and FastAPI for low-
latency inference, and implemented monitoring and alerting with MLflow and Great Expectations to track
model performance and data drift. - SScalable ML pipeline for predictive maintenanceImplemented a scalable ML pipeline for predictive maintenance using IoT sensor data stored in AWS S3 Data
Lake. Preprocessed high-volume streaming data with PySpark and Spark Structured Streaming, built LSTM
and Random Forest models for early anomaly detection, and deployed solutions using Docker, Kubernetes,
and AWS SageMaker for production-ready predictive maintenance applications. - EEnd-to-end ML pipelines for customer churn predictionDesigned and developed end-to-end ML pipelines for customer churn prediction using Python, Pandas, Scikit-
learn, XGBoost, and LightGBM. Processed large-scale datasets with PySpark and SQL, performed feature
engineering, model evaluation, and hyperparameter tuning, deployed models using FastAPI, Docker, and
AWS SageMaker, and created interactive dashboards with Tableau and Power BI for churn risk monitoring.