JamilHasanRahul

• ›Architected and shipped ValorPro - a 10-phase multi-agent voice interview system for US veteran VA disability claims on VAPI, FastAPI, PostgreSQL, Redis, and Docker.
• ›Built an end-to-end AI recruitment pipeline automating 90% of manual onboarding tasks and eliminating ghost access risks through centralised deprovisioning.
• ›Developed Monica, a city-aware concierge AI with geo-filtered KB retrieval deployed in production for a Romanian enterprise client.
• ›Built an AI-powered invoice OCR pipeline with dual-database QA→Production migration enforcing a ≥90% correctness threshold.

• ›Fine-tuned image classification and object detection models using TensorFlow/Keras and PyTorch.
• ›Built augmentation pipelines to improve model robustness and generalisation.
• ›Integrated CV models into internal evaluation pipelines; benchmarked and optimised inference speed.

• ›Produced SEO-optimised technical articles on software, hardware, and consumer technology.
• ›Maintained and developed front-end components in HTML, CSS, and JavaScript.

Priority-aware resource allocation system for IoMT networks, integrating device-to-device offloading and MEC architecture. Proposed an MILP optimisation framework targeting latency and energy efficiency.

Exploring ML approaches - Random Forest, LLM, GTrXL - on sensor data for student stress and mental health detection.

Ensemble Learning model for forecasting air quality across major divisions of Bangladesh.