Crafting ML systems from data to delightful products.
Notes, experiments, and production lessons on time-series, NLP, LLMs, and scalable MLOps.
Machine Learning • Transformers
Fake PDF Detection Using Transformer
LLM • Few-Shot Learning
Few-Shot Intent Classification in Conversational AI with Prompt-Tuned LLMs
Climate Modeling • GANs
Climate Data Downscaling Using Conditional GANs
LLM • Code Review
Automatic Code Review Comment Generation Using Large Language Models
NLP • Adversarial Attacks
Adversarial Attack Detection in NLP Models Using Statistical Outlier Analysis
Machine Learning • Data Science
The Importance of Machine Learning and Data Science in Terms of Data Modeling
Latest
View allMachine Learning • Transformers • Document Forensics
Fake PDF Detection Using Transformer
A transformer-based approach to detecting fraudulent PDF documents using natural language processing, metadata, and structure analysis.
LLM • Few-Shot Learning • Conversational AI
Few-Shot Intent Classification in Conversational AI with Prompt-Tuned LLMs
Prompt-tuning strategies for low-data intent classification across domains.
Climate Modeling • GANs • Deep Learning
Climate Data Downscaling Using Conditional GANs
Learning high-resolution climate fields from coarse simulations with cGANs.
LLM • Code Review • Software Engineering
Automatic Code Review Comment Generation Using Large Language Models
LLM-driven, style- and semantics-aware suggestions for code review workflows.
NLP • Adversarial Attacks • Outlier Detection
Adversarial Attack Detection in NLP Models Using Statistical Outlier Analysis
Detecting adversarial text inputs via distributional shift and influence diagnostics.
Machine Learning • Data Science • Data Modeling
The Importance of Machine Learning and Data Science in Terms of Data Modeling
## Abstract Machine Learning (ML) and Data Science (DS) have become inseparable components of modern data-driven innovation. While ML provides the algorithms an
Machine Translation • Transformers • Transfer Learning
Low-Resource Machine Translation with Transformer Models and Transfer Learning
Leveraging multilingual pretraining and adapters for low-resource translation.
Finance • LSTM • XGBoost
Financial Time-Series Forecasting Using Hybrid LSTM–XGBoost Models
A hybrid pipeline that pairs sequence encoders with tree-based residual learners for market forecasting.
Explainable AI • Vision Transformers • Medical Imaging
Explainable AI for Medical Imaging Diagnosis Using Grad-CAM and Vision Transformers
Combining ViT with gradient-based attribution for faithful explanations in radiology.
Transformers • Emotion Recognition • Speech Processing
Multimodal Emotion Recognition Combining Speech and Facial Cues via Transformers
A transformer that fuses acoustic and visual streams for robust emotion recognition.
IoT • LSTM • Time Series
Real-Time Anomaly Detection in IoT Sensor Data Using LSTM Networks
Detecting anomalies in streaming IoT telemetry using sequence models and robust thresholds.