Work
Turning data, models, and ideas into things that actually matter — from ML pipelines processing 1M+ transactions to AI systems rethinking how teams work.
Converts graph data (nodes and edges) into fluent natural language using graph-to-text techniques and NLP — bridging structured data and human-readable insight.
End-to-end ML pipeline predicting CLV using RFM features, XGBoost, and K-Means clustering on 1M+ transaction records — production-grade data science at scale.
Product case study proposing a collaborative multi-user AI system — redesigning from individual research to team intelligence with feature-level proposals to improve retention and productivity.
Real-time detection and tracking using YOLOv8 and SORT with Kalman filtering on drone imagery — designed for robust performance in dynamic aerial environments.
Transfer learning on ResNet-50 for skin lesion classification — leveraging pre-trained features to achieve high accuracy on imbalanced medical imaging datasets.
Identifies key influencers in a network using graph-based analysis and centrality metrics — PageRank, betweenness, and eigenvector centrality on real network data.
ML model detecting phishing and malicious URLs using lexical and domain-based features — real-time threat identification with high precision and recall.
Analytics platform for AQI and pollution trends using time-series insights and interactive visualization — surfacing actionable environmental intelligence.
Transformer-based multilingual translation system powered by HuggingFace with real-time inference — fine-tuned for domain-specific language pairs.
CNN-based pneumonia detection using MobileNetV2 with uncertainty estimation — interpretable clinical AI with confidence calibration for medical decision support.
Personal finance platform for expense tracking, budgeting, and financial insights — a clean, intuitive dashboard UI that makes money management effortless.
Healthcare platform for managing patient data and improving doctor-patient connectivity — streamlining clinical workflows and reducing administrative overhead.
Structured product requirements document covering problem definition, user personas, feature prioritization, and a phased roadmap — rigorous PM thinking in practice.
I'm available for collaborations, consulting, and full-time roles.