Full-Stack Software Engineer, IT Consultant & AI Researcher
I build scalable software for startups and enterprises. Whether you need an AI-powered SaaS platform, a custom CRM, HRM, POS system, or a web application, I take care of the full lifecycle: consultation, architecture, development, deployment, and ongoing maintenance.
Strategic technical planning, system architecture design, technology stack evaluation, and roadmap development tailored to your business goals.
Discuss Your ProjectEnd-to-end development of SaaS platforms, CRM, HRM, POS systems, web applications, and custom enterprise software using modern tech stacks.
Start BuildingOptimizing AWS/Azure infrastructure for cost reduction, security auditing, and performance tuning. Cloud migration and automated compliance.
Optimize CostsBuilding robust RESTful APIs with clear documentation, third-party integrations, payment gateways, and microservice communication design.
Integrate SystemsImplementing AI-powered features, computer vision systems, predictive models, and machine learning pipelines for intelligent applications.
Explore AIProduction deployment, CI/CD pipelines, Docker containerization, server management, and ongoing support for live applications.
Deploy Now
Python & AWS
Absolute Ops, a FinOps platform optimizes cloud workloads for performance and cost, with built-in security analysis.
SaaS • Django
Real estate intelligence platform for U.S. investors to select market and property to invest in best possible way.
Microservices
Microservices-based backend system for managing user cards, integrating high-speed scraping and Typesense search.
An in-house HRM system for managing Mediusware LTD. employees, payroll, and attendance.
Mediusware Ltd.
Developed and maintained web applications, implemented new features, and optimized performance for client projects.
Mediusware Ltd.
Built responsive web interfaces, collaborated with design team, and contributed to agile development processes.
This paper compares Vector AutoRegression (VAR), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) models for forecasting multivariate IMU-based gait data to monitor Parkinson's Disease, finding GRU to be the most accurate model.
This paper proposes a privacy-preserving federated learning framework with a dual-stream ResNetRS50 backbone for multi-scale colorectal cancer histopathological grading, achieving 83.5% accuracy and high recall (87.5%) for the critical Grade III tumors.
View on arXivThis study proposes an interpretable deep learning framework using an EfficientNet-B2 backbone with a prototype-based head and Out-of-Distribution (OOD) detection for multi-class tea leaf disease classification, achieving a balanced accuracy of 97.87% on the Tea LeafBD dataset.
Whether you are a startup building your first MVP or an established company that needs scalable infrastructure, I can help you ship reliable software that delivers real results.