I designs the system architecture, builds the ML and backend components, ships them to production on the cloud, and stays on to maintain and improve the builds the entire lifecycle, under one engineer.
Most AI projects fail between the handoffs — a model that never gets deployed, an API nobody maintains. Anand owns the whole path from architecture to production.
Anand works as an independent full-stack AI & cloud engineer, taking projects from problem formulation through model development, backend integration, and production deployment — then staying on to keep them running.
His background spans backend and cloud engineering — microservices, caching, CI/CD, and secure API design — alongside applied machine learning, including RAG pipelines, Neural ODEs, and deep learning classifiers. He also builds real-time data systems like live dashboards and streaming architectures.
He holds a B.Tech in Computer Science and Engineering and is an EC-Council Certified Ethical Hacker (CEH v12), bringing a security-conscious approach to every system he ships.
Open-source, shipped, and verifiable — every project below links straight to the code.
Real-time disaster intelligence dashboard that tracks global calamity events on an interactive 3D globe, streams high-severity flash alerts over WebSockets, and supports geospatial search with resilient event caching.
Full-stack · live in productionAn encoder–Neural ODE–decoder framework learning continuous latent dynamics for predictions at arbitrary, irregular timestamps — tested on damped-oscillator, Van der Pol, and Lotka–Volterra systems.
Validation MSE: 1.831 → 0.028End-to-end deep learning pipeline for high-dimensional gene-expression classification, with configurable MLPs, stratified k-fold CV, and support for Golub ALL/AML and TCGA PANCAN datasets.
Biomedical, limited-sample robustDeep learning framework for DNA splice-site classification — exon–intron, intron–exon and non-splice sequences — benchmarking CNN, BiLSTM, and Transformer architectures.
96.6% accuracy · 0.961 macro-F1Inventory analytics for identifying slow-moving stock, estimating holding-cost impact, and supporting replenishment and discount decisions via a Random Forest workflow.
99.49% accuracy · 0.99 F1Reusable manufacturing analytics component for quality monitoring, defect analysis, machine/shift comparison, KPI computation, and production trend visualization.
Built for integration into larger systemsCurrently taking on freelance and contract work — from system design through cloud deployment and long-term updates.