Aman Kumar

Software Engineer

• Led backend development for all Vision tools at Ripik.ai, utilizing technologies such as Java (Spring Boot), Python (Django, Flask), and Computer Vision-based AI systems.
• Streamlined Vision use cases by aligning tool functionalities into a unified flow, allowing integration into a common dashboard. Contributed to codebase migration from Django/Flask to Spring Boot with MongoDB for better maintainability and consistency.
• Identified and resolved critical system vulnerabilities, significantly improving the scalability, reliability, and deployment efficiency of Vision services across environments.
• Collaborated with Data Science teams to deploy AI/ML models on both cloud platforms (AWS, Azure) and edge/client-side environments, ensuring flexible and robust model serving.
• Instituted clean coding and DevOps practices, including the development of a centralized Python package repository. Hands-on experience with Docker, AWS (EC2, S3, EBS, ECS), and implementing CI/CD pipelines using GitHub Actions.
• Designed, Built, and Deployed: AI Sandbox platform from scratch, enabling customers to test their ML models and integrate Azure Custom Vision, using Java, Spring Boot, Python, AWS, Azure, and PostgreSQL.

Software Engineer

• Worked as a Software Developer in the Credit team, building and optimizing financial microservices using Java, Spring Boot, MySQL.
• Designed, developed, and maintained RESTful APIs that powered critical Credit Loan and Load Collection workflows, enabling efficient transaction processing for end-users.
• Proactively resolved Insecure Direct Object Reference (IDOR) vulnerabilities in multiple APIs, enhancing the platform’s security and ensuring compliance with industry regulations.

Software Engineer

• Diagnosed and Resolved production issues in the core product by performing in-depth debugging across frontend (Vue.js, JavaScript) and backend (Java) components, ensuring uninterrupted business operations.
• Addressed UI bugs and performance concerns in collaboration with QA and Design teams, providing detailed Root Cause Analysis (RCA) to support continuous improvement and knowledge sharing.
• Contributed to Stability and Performance improvements within the Analytics and Storage modules by fixing critical issues, supporting high-intensity data-driven features.

Software Engineering Intern

• Developed automation scripts for generating Solar Output reports using PVSyst software, leveraging Python automation libraries such as pyautogui and pywinauto.
• Built RESTful APIs to accept JSON input data and trigger backend automation workflows, utilizing Flask along with desktop automation tools in Python.
• Designed and implemented MySQL database schemas for structured logging of API requests, configured with Master-Slave replication to ensure high availability and performance.
• Automated the generation of response files and integrated seamless upload workflows to cloud storage using Python and AWS services.

• Reimplemented Solar Access generation logic originally written in Python into optimized C and C++ modules for enhanced performance and portability.
• Refactored NumPy-heavy Python scripts into C++ by designing efficient class-based structures for matrix operations, significantly reducing computational overhead.
• Utilized modern C++ JSON parsing libraries such as RapidJSON and nlohmann/json to handle complex data interchange formats with minimal latency.
• Enabled high-performance parallel processing in computationally intensive modules by developing CUDA-based submodules using the NVCC CUDA platform.

• Developed RESTful APIs to simulate Google Maps API functionality for integration into The Design Studio, a flagship product by The Solar Labs.
• Engineered backend automation to retrieve satellite imagery by controlling Google Earth software, based on input parameters like latitude, longitude, and zoom level — detailed in the GEP Text Extractor module.
• Enabled cloud-based delivery of historic satellite images critical for solar rooftop design by overcoming cloud coverage issues, uploading processed data to cloud storage, and returning access links through APIs — implemented using Flask, Python, AWS, GCP, and Google Earth. Related module: Image to JSON Encoder.

Indian Institute of Information Technology, Nagpur