You most probably have an issue with the background. Have you obtained the same results when repeating the measurements? Have you collected prior to your measurements the background spectrum (on clean ATR)?

Also, there are several discussions here on RG on the topic of IR inverted peaks that I’m sure will help you.

https://www.researchgate.net/post/What_is_the_cause_for_FTIR-ATR_inverted_peaks_in_otherwise_normal_spectrum

https://www.researchgate.net/post/How-to-explain-the-presentation-of-inverted-absorption-bands-in-an-FTIR-spectrum

Following up on Ioana Maria Cortea 's answer, the most noticeable reversed peaks are those of water if I am not mistaken. Thus I think the problem comes from your background that is too old and does not represent your current 'blank' when measuring your spectrum. When doing ATR I used to measure a new background prior to each sample measurement, here it looks like you had more water before your actual spectrum. Moreover, you have a quite low transmittance %, it can explain why the inverted peaks seem so intense, relatively to your sample peaks. Hope this helps !

Benjamin Martial sir I take background in ATR before doing every sample

First of all, I am somewhat puzzled that you have transmittance on the ordinate while it should be reflectance. Anyway, to better understand the problem, how do you record your background? Standard would be to measure the spectrum without any sample - many problems arise because for the background a kind of reference sample is chosen instead. How is it in your case?