Raylib's suggest is like this.

saveandload.cpp

// Save integer value to storage file (to defined position)
// NOTE: Storage positions is directly related to file memory layout (4 bytes each integer)
bool SaveStorageValue(unsigned int position, float value, const char* _fileName)
{
    bool success = false;
    unsigned int dataSize = 0;
    unsigned int newDataSize = 0;
    unsigned char *fileData = LoadFileData(_fileName, &dataSize);
    unsigned char *newFileData = NULL;
 
    if (fileData != NULL)
    {
        if (dataSize <= (position*sizeof(float)))
        {
            // Increase data size up to position and store value
            newDataSize = (position + 1)*sizeof(float);
            newFileData = (unsigned char *)RL_REALLOC(fileData, newDataSize);
 
            if (newFileData != NULL)
            {
                // RL_REALLOC succeded
                float *dataPtr = (float *)newFileData;
                dataPtr[position] = value;
            }
            else
            {
                // RL_REALLOC failed
                TraceLog(LOG_WARNING, "FILEIO: [%s] Failed to realloc data (%u), position in bytes (%u) bigger than actual file size", *_fileName, dataSize, position*sizeof(int));
 
                // We store the old size of the file
                newFileData = fileData;
                newDataSize = dataSize;
            }
        }
        else
        {
            // Store the old size of the file
            newFileData = fileData;
            newDataSize = dataSize;
 
            // Replace value on selected position
            float *dataPtr = (float *)newFileData;
            dataPtr[position] = value;
        }
 
        success = SaveFileData(_fileName, newFileData, newDataSize);
        RL_FREE(newFileData);
 
        TraceLog(LOG_INFO, "FILEIO: [%s] Saved storage value: %i", _fileName, value);
    }
    else
    {
        TraceLog(LOG_INFO, "FILEIO: [%s] File created successfully", _fileName);
 
        dataSize = (position + 1)*sizeof(float);
        fileData = (unsigned char *)RL_MALLOC(dataSize);
        float *dataPtr = (float *)fileData;
        dataPtr[position] = value;
 
        success = SaveFileData(_fileName, fileData, dataSize);
        UnloadFileData(fileData);
 
        TraceLog(LOG_INFO, "FILEIO: [%s] Saved storage value: %i", _fileName, value);
    }
 
    return success;
}
 
// Load float value from storage file (from defined position)
// NOTE: If requested position could not be found, value 0 is returned
float LoadStorageValue(unsigned int position, const char* _fileName)
{
    float value = 0;
    unsigned int dataSize = 0;
    unsigned char *fileData = LoadFileData(_fileName, &dataSize);
 
    if (fileData != NULL)
    {
        if (dataSize < (position*sizeof(float))) TraceLog(LOG_WARNING, "FILEIO: [%s] Failed to find storage position: %i", _fileName, position);
        else
        {
            float *dataPtr = (float *)fileData;
            value = dataPtr[position];
        }
 
        UnloadFileData(fileData);
 
       TraceLog(LOG_INFO, "FILEIO: [%s] Loaded storage value: %f", _fileName, value);
    }
 
    return value;
}

But by this way, saved file is binary. I don't like this binary file.