Moazzam irshad

I’m a Data Science undergraduate with hands-on, end-to-end experience building and deploying machine learning systems—from recommendation engines and multi-class clinical classifiers to real-time edge AI computer vision pipelines. I’m comfortable across the full ML workflow: data cleaning, feature engineering, model selection, cross-validation, and deployment on constrained hardware.

I focus on practical problem diagnosis at the systems level. In my edge AI work, I resolved low-level GStreamer/OpenCV and TensorRT build conflicts—so hardware-accelerated video capture and inference work reliably for live pipelines on NVIDIA Jetson Nano.

On the modeling side, I’ve built production-minded classifiers and evaluation workflows. For example, I developed a 3-class liver cirrhosis severity classifier using forward-fill imputation, ordinal/binary encoding, log-transformed biomarkers, and weighted F1 scoring across 5-fold cross-validation, selecting a production candidate based on generalization behavior.

I also ship complete end-to-end solutions: a dual-model content recommendation engine (content-based TF-IDF/cosine similarity plus clustering with KMeans) and real-time object detection pipelines (CPU OpenCV DNN + MobileNet-SSD and a GPU-accelerated TensorRT pipeline). I’m now looking for an entry-level role to apply these skills in a production ML/data environment.