Designed anomaly detection models using deep learning on network telemetry data, improving true-positive detection rates by 32% and reducing false positives. Implemented Transformer-based NLP solutions (BERT, GPT) for automated log parsing, reducing mean-time-to-detect incidents. Built streaming feature pipelines with Kafka and Spark for real-time analysis of over 5 million daily network events. Deployed ML models on Kubernetes with Docker and MLflow, reducing inference latency from 280ms to 110ms and increasing service uptime. Developed a retrieval-augmented generation (RAG) assistant with LangChain and vector databases, cutting troubleshooting time for support engineers. Integrated AI-based intrusion detection with Splunk and QRadar, enhancing security monitoring coverage and SOC efficiency.

Machine Learning Engineer at Datamatics

July 1, 2021 - July 1, 2023

Delivered fraud detection models for financial transactions using XGBoost and ensemble methods, achieving 96% precision and reducing false investigations. Developed and deployed recommendation systems on AWS SageMaker, driving a 20% increase in customer engagement and 8% higher conversions. Applied CNN and Transformer-based OCR models for document automation, reducing manual data entry effort by 45%. Automated ML workflows with Airflow and Spark on Databricks, reducing ETL and feature engineering time by 20%. Established CI/CD pipelines with GitHub Actions and MLflow, enabling safe weekly releases with blue-green deployments. Built real-time monitoring dashboards with Prometheus and Grafana for drift detection and SLA tracking, reducing downtime and data quality issues.

Data Scientist / AI Engineer at 3i Infotech

January 1, 2019 - June 1, 2020

Built predictive models for customer segmentation and churn reduction, increasing campaign ROI by 10%. Implemented NLP sentiment and topic models with Word2Vec and LSTM to analyze customer feedback, reducing recurring complaints. Automated ETL processes using Python and SQL with data validation checks, improving reporting efficiency by 20%. Designed interactive Tableau dashboards to track ML model performance and business KPIs, improving executive decision-making speed. Deployed AI models in production with Docker and versioned training environments, streamlining delivery to IT teams. Established feature reuse across multiple projects with a centralized feature store, improving productivity and consistency. Monitored schema drift and implemented safeguards against null-value issues, reducing data-related outages.