Olivia Chatterjee

Designed and implemented advanced predictive models using machine learning and deep learning (e.g., LSTM, DRL) for multi-horizon vulnerability estimation.

Collaborated with interdisciplinary teams to integrate diverse country-specific data streams into robust modeling pipelines, enhancing data richness and applicability.

Validated and refined models through rigorous historical analysis, improving accuracy and robustness for early warning capabilities and geopolitical risk analysis.

Achieved ~75% classification accuracy on cancer datasets using transformed tabular genomic data with CNNs, demonstrating deep learning feasibility for structured biomedical data.

Developed and implemented a novel method to convert tabular data into image-like structures for CNN processing, addressing key limitations in prior studies and expanding CNN applications.

Conducted in-depth analysis of existing research and methodologies in tabular data classification using CNNs, identifying challenges and establishing a generalizable pipeline for disease detection.

Achieved over 99% model accuracy in predicting fetal risks of congenital disorders using neural networks and ensemble learning, even after training on privatized data.

Implemented Data De-identification and Privatization using Differential Privacy, ensuring secure and cost-effective prenatal risk screening while safeguarding sensitive patient data.

Developed and fine-tuned artificial neural network models, regression models, and ensemble methods (XGBoost, Random Forest) for accurate risk prediction.

Deployed the AI Engine API into production, reducing dependence on expensive confirmatory diagnostics and establishing an operational, privacy-preserving healthcare solution.

Analyzed pitch decks of leading startups and delivered investment summaries with market insights, gaining practical exposure to VC decision-making processes.

Conducted comprehensive market research and competitive analysis on emerging energy segments, synthesizing insights into actionable investment summaries.

Developed critical skills in early-stage innovation assessment within the clean-tech space, understanding investment theses and venture capital fundamentals.

Developed machine learning models capable of identifying early indicators of market crashes through pattern detection and classification, analyzing a decade-long stock price dataset.

Implemented and compared various supervised learning algorithms, optimizing models by experimenting with different parameters to classify market conditions as crash-prone or stable.

Conducted exploratory data analysis to identify key trends and patterns, ensuring data integrity by handling missing data and creating a feature matrix from consecutive Close prices.

Evaluated classification algorithm performance using metrics like confusion matrices and classification reports, iteratively improving prediction accuracy and laying groundwork for financial market anomaly detection.

Engineered classification logic using Prolog to detect hemoglobinopathies based on retention times and peak structures, automating identification of known and unknown peaks.

Developed and applied Python-based encryption techniques to secure sensitive medical data, ensuring patient information privacy and protection throughout the diagnostic process.

Successfully deployed the platform in HPLC machines across hospitals and diagnostic labs, significantly reducing diagnostic turnaround times and establishing a scalable, privacy-preserving solution.

Automated interpretation of liquid chromatogram data for HbA1c analysis, enabling rapid classification of hemoglobin variants and reducing manual effort.