Aryan Anand

Ninestars Information Technologies, Bangalore

• Architected the backend of AOTM FDI, a multi-tenant credential-fraud detection platform with 11 Django apps, 119 REST routes and a 43-table PostgreSQL schema; built a 5-stage OCR-to-clustering pipeline using RabbitMQ quorum queues and specialised Celery workers behind a Uvicorn ASGI application.
• Designed reliable distributed processing with idempotent writes, fan-out/fan-in orchestration, transactional coordination and presigned object storage access; enforced fail-closed multi-tenancy across 33 tenant-scoped tables with RBAC, rotating JWT sessions and encrypted exports, shipping 170+ versioned builds in seven weeks backed by 70 scripted test suites.
• Built a multi-tenant rate-limiting and usage-based billing system using Redis Lua admission, hierarchical limits, PostgreSQL-backed credit reservations, cache invalidation and background reconciliation, delivering a sub-5ms hot path with zero PostgreSQL queries on cache hits.
• Engineered a distributed database migration engine for parallel extraction and direct-to-S3 upload of 60M+ row databases, with daemon-based processing, live progress reporting, dynamic concurrency and graceful cancellation; also built REST/GraphQL APIs, WebSocket progress streams and optimistic-locked collaborative workflows.
• Built distributed ML processing across 10 RabbitMQ quorum queues with transactional fan-in and redelivery-safe processing, executing 10,700+ inference jobs across 2,500+ completed multi-stage runs; operated Docker-based ASGI/Celery services with Jenkins CI and automated database, cache, broker and storage provisioning.

TIH IoT, IIT Bombay, Mumbai

• Built a U2-Net pipeline for automated image masking and applied CNNs on NDVI and hyperspectral imagery for vegetation segmentation (11% gain), preprocessing, augmentation, and hyperparameter tuning for deployment.

Python, PyTorch, TensorFlow, Llama, React.js, Node.js, MongoDB, Git

• Built a super-resolution GAN (TensorFlow) for MRI/CT clarity achieving 92% accuracy on 54,000 samples, and a CycleGAN (PyTorch) for bidirectional MRI-to-CT translation; designed a hybrid Vision Transformer-CNN for multi-label disease classification with 88.4% accuracy.

Python, OpenCV, TensorFlow, Google Cloud, Git

• Developed a real-time hand sign language-to-speech system using ResNet-18 and OpenCV achieving 97% accuracy and 200ms latency on low-end devices, with Google TTS API integration.