The continuously increasing number of Wireless Fidelity (WiFi) access points have resulted in increased congestion in already congested indoor networks. In recent years, Light Fidelity (LiFi) has presented itself as upcoming access technology to complement WiFi owing to its vast spectrum, high data rates, and secrecy. Infrared (IR) is widely adopted in the uplink of LiFi-based systems. Due to the limited availability of energy in the Internet of Things (IoT) nodes, energy efficiency is vital. This paper presents a client-based adaptive IR uplink transmit power algorithm to enhance energy efficiency while satisfying the data rate requirement of IoT nodes. With hardware-based experimentation with LZ1 IR-LED and Heidi lens, the need for such an adaptive transmit power mechanism is first demonstrated. Simulation-based evaluation is then performed, showing that the proposed algorithm outperforms random and fixed uplink optical powers in terms of energy efficiency and satisfaction of data rate requirements of IoT nodes. The results clearly show the need for hardware design of IR-based IoT nodes with integrated adaptive uplink transmit power algorithms for LiFi IoT networks.