Portfolio
A curated collection of ML systems, data pipelines, and engineering projects — each with a full case study covering architecture, outcomes, and tech stack.
Filter by skill
End-to-end training & inference at scale
A production-grade distributed training pipeline built on PyTorch and Ray, capable of orchestrating multi-GPU workloads across heterogeneous clusters.
Retrieval-augmented generation for enterprise docs
A retrieval-augmented generation system that indexes internal documentation and surfaces precise, cited answers using a fine-tuned LLM.
YOLO-based detection pipeline for edge devices
A computer vision pipeline that runs YOLOv8 inference on edge hardware, streaming annotated video to a central dashboard with sub-30ms latency.
Unified storage & feature engineering platform
A Delta Lake-based data lakehouse that unifies raw data ingestion, feature engineering, and model training datasets under a single governance layer.
This site — Next.js 14 + MDX + contentlayer
A performant, statically generated portfolio built with Next.js 14 App Router, MDX-powered project case studies, and a custom animated logo matrix.
Systematic evaluation of AutoML frameworks
A research-grade benchmarking harness that evaluates AutoML frameworks (AutoGluon, H2O, FLAML) across 50+ tabular datasets with reproducible metrics.
Live GitHub Integration — This page is dynamically populated with data from my GitHub profile, including project descriptions and languages for filtering.