https://obadaalsehli.com Technical writing on image processing, AI systems, and reinforcement learning by Obada Alsehli. en-US https://obadaalsehli.com/writing/liveness-is-not-model-readiness https://obadaalsehli.com/writing/liveness-is-not-model-readiness What separate liveness, health, and readiness responses mean in a documented vision-model API, and what they do not establish about prediction quality. https://obadaalsehli.com/writing/silver-labels-are-not-ground-truth https://obadaalsehli.com/writing/silver-labels-are-not-ground-truth What do multi-head vision benchmarks measure when annotations come from heuristics and VLMs? This write-up examines the four-head classification setup, reported F1 metrics, and a separate 210-image human label audit. https://obadaalsehli.com/writing/degenerate-policy-in-a-highway-rl-agent https://obadaalsehli.com/writing/degenerate-policy-in-a-highway-rl-agent A PPO agent learned to favor slow driving rather than the intended behavior. This write-up examines the reported training results and reward setup. https://obadaalsehli.com/writing/leap-2026-privacy-and-memory https://obadaalsehli.com/writing/leap-2026-privacy-and-memory A reflection from the Syrian pavilion at LEAP 2026 on the engineering questions behind privacy-first AI, conversational memory, and user control. https://obadaalsehli.com/writing/agent-driven-development-product-judgment https://obadaalsehli.com/writing/agent-driven-development-product-judgment An opinion on why domain understanding, explicit outputs, and independent validation remain central when coding agents accelerate implementation. https://obadaalsehli.com/writing/leap-2026-healthcare-ai-workflows https://obadaalsehli.com/writing/leap-2026-healthcare-ai-workflows A field note from LEAP 2026 on the different workflows behind healthcare AI—and why the task matters more than the AI-powered label. https://obadaalsehli.com/writing/leap-2026-managed-rag-orchestration https://obadaalsehli.com/writing/leap-2026-managed-rag-orchestration An AWS AI Training Academy reflection on choosing ownership boundaries for retrieval, orchestration, and sensitive data instead of treating managed versus custom as a binary. https://obadaalsehli.com/writing/ddia-learning-the-why https://obadaalsehli.com/writing/ddia-learning-the-why A personal reading reflection on moving from following implementation steps to explaining data-system decisions and trade-offs. https://obadaalsehli.com/writing/learning-deep-learning-by-explaining https://obadaalsehli.com/writing/learning-deep-learning-by-explaining A personal reflection on using equations, explanations, comparisons, and retrieval questions to build a reusable learning practice around a deep-learning course. https://obadaalsehli.com/writing/bau-workshop-architecture-and-specifications https://obadaalsehli.com/writing/bau-workshop-architecture-and-specifications A reflection on a BAU team workshop: tracing an AI-enabled application across layers and connecting architecture, specifications, and independent validation. https://obadaalsehli.com/writing/idms-capstone-shared-contracts https://obadaalsehli.com/writing/idms-capstone-shared-contracts A team reflection from the IDMS capstone presentation on shared schemas, software–AI handoffs, and credit across a six-person project. https://obadaalsehli.com/writing/agent-browser-is-an-interface https://obadaalsehli.com/writing/agent-browser-is-an-interface Browser choice, model context and session persistence solve different problems. Before replacing an automation tool, identify what the task needs—and what evidence is missing.