An MLOps workflow for training, inference, experiment tracking, model registry, and deployment.

Computes CT contrast phase and GI tract contrast using TotalSegmentator and ML

A comprehensive .NET MAUI plugin for ML inference with ONNX Runtime, CoreML, and platform-native acceleration support

gRPC server for Machine Learning (ML) Model Inference in Rust.

EcoChain-ML is a hybrid energy-aware ML framework integrating a lightweight PoS blockchain layer and renewable-aware scheduling. Built to simulate green computing strategies on a single PC, it evaluates energy, latency, and sustainability trade-offs.

[TPDS 2025] EdgeAIBus: AI-driven Joint Container Management and Model Selection Framework for Heterogeneous Edge Computing

Dockerized Django application for handwritten math expression recognition using a CNN model, with end-to-end ML pipeline and cloud-ready deployment.

Production-style real-time ML feature store with low-latency inference

ML service for cats that actually learn stuff. PPO brains, personality drift, mood system.

🐱 Create a living cat AI that exhibits emotions, reactions, and realistic behavior for an engaging and interactive experience.

PoC demonstrating distributed workload orchestration using Ray as the primary compute framework with Prefect for workflow orchestration, supporting cloud-native deployments (Kubernetes)

AI recruitment intelligence platform with resume scoring, role matching, and inference workflow design.

High-performance C++20 neural network framework powered by Intel oneAPI MKL 2025.2. Optimized for CPU-based deep learning inference and training.

Containerized ML inference service exposing a churn prediction model via FastAPI, with Docker-based deployment and AWS-ready architecture.

Enterprise Data Warehouse & ML Platform - High-performance platform processing 24B records with <60s latency and 100K records/sec throughput, featuring 32 fact tables, 128 dimensions, and automated ML pipelines achieving 91.2% accuracy. Real-time ML inference serving 300K+ predictions/hour with ensemble models.

Submission of Project

Client-side React + Vite web app to record and process voice audio and send features to an API for automated stress prediction (speech-based stress detection).

A lightweight, framework-agnostic middleware that dynamically batches inference requests in real time to maximize GPU/TPU utilization.

CNN-based vehicle number plate detection and recognition system using TensorFlow, OpenCV, and Flask API for real-time image inference and automated license plate identification

Production-grade AutoPrep AI ML Data Platform for automated data preprocessing, AutoML training, feature store management, vector search, and real-time inference.