Sign language is the primary mode of communication for Hearing and Speech Impaired (HSI) people. However, the complexity and intricate nature of Indian Sign Language, which includes a majority of double-handed signs, poses a challenge for HSI people to communicate effectively with others. Moreover, the expanding vocabulary of sign language makes it difficult for those without access to updates to communicate effectively. Fingerspelling is most widely used by the HSI for general and easy day-to-day communication. A real-time and efficient fingerspelling system is thus crucial to facilitate communication for HSI people in a natural setting. However, existing real-time recognition systems are cumbersome and inefficient as they employ complex deep-learning architectures and primarily use RGB image and video data that are sensitive to lighting and background conditions and therefore are more error prone and moreover do not perform well under natural settings. This study proposes a simple and efficient real-time fingerspelling system for recognizing static fingerspelling gestures using Leap Motion Controller. The study employs a random forest classifier with translation-independent features to recognize signs, while achieving comparable accuracy, making the overall system lightweight. We achieved a real-time validation accuracy of 71% while also predicting the sample instantaneously with an average response time of 3.02 ms. Since fingerspelling can introduce spurious signs during transitions and can be ambiguous when recognizing similar signs, our system also includes a word fine-tuning phase that uses a dictionary-based approach to simplify the recognition process, making our system well-suited for real-time deployment in natural settings.