Overview
At Cmind AI, I work as an Associate Data Scientist focusing on building production-ready machine learning systems for financial applications. My work spans from architecting ML microservices to developing end-to-end NLP pipelines.
Key Projects
ML Microservices Architecture
Architected and deployed modular machine-learning microservices for feature selection, class imbalance handling, and model training. This microservices approach enables independent scaling and significantly reduces production failure risk by isolating components.
Technologies: Python, Docker, REST APIs, Cloud Infrastructure
Earnings Per Share (EPS) Surprise Prediction System
Enhanced the company’s flagship EPS Surprise Prediction system by implementing automated retraining pipelines. I trained and evaluated multiple model architectures including:
- XGBoost - For gradient boosting on structured financial data
- Support Vector Machines (SVM) - For classification tasks
- Recurrent Neural Networks (RNN) - For capturing temporal patterns in sequential data
The system processes 15+ years of historical financial data and achieves 90% quarterly forecast accuracy.
Sentiment Analysis Pipeline
Led development of a customizable end-to-end sentiment-analysis pipeline featuring:
- FinBERT integration - Fine-tuned BERT model for financial domain text
- OpenAI APIs - For tailored Q&A with explanations
- Oracle Cloud ingestion - For cost-efficient legacy-system compatibility
- AWS S3 storage - Supporting smooth migration to modern analytics workflows
MLOps & Experiment Tracking
Leveraged MLflow throughout all projects for:
- Experiment tracking and comparison
- Model version control
- Deployment monitoring
- Ensuring reproducibility and compliance with internal MLOps standards
Skills Developed
- Feature Selection & Engineering
- Model Training & Evaluation
- Class Imbalance Handling
- MLOps Best Practices
- Financial Domain NLP
- Cloud Infrastructure (AWS, Oracle)