In Tanakh, the name Hillel is always found defective: הִלֵּל (Judges 12:13, etc.). This form is also found in the Bar-Kosiba letters recovered at Wadi Murabbaʿat (הלל בן גריס, Mur. 24).
However, the name Hillel is found both plene or defective in the best rabbinical manuscripts. Take, for example, the Codex Kaufmann A50, widely considered to be (one of) the most exact Mishah manuscripts. There, one finds both הֶלֵּל and הֶילֵּל, even in close proximity. In Berakhot 8:1, we find the following:
... 'אֵלּוּ דְבָרִים שֶׁבֵּין בֵּית שַׁמַּיִ לְבֵית הֶלֵּל בִּסְעוֹדָה בֵּית שַׁמַּיִ אוֹמְרִין מְבָרֵךְ עַל הַיּוֹם וְאַחַר כָּךְ מְבָרֵךְ עַל הַיַּיִן [וּ]בֵית הֶילֵּל אוֹמְ
MS Parma A and MS Add. 470.1 give both names as defective in this verse.
The Leiden MS is the only complete manuscript of Yerushalmi is extant, and it too has both הלל and הילל forms. See Shevi'it 10:3:
מה שתיקן הילל פרוזבול. ברם כמן דא. מעשרות מדבר תורה. והלל מתקין על דבר תורה.
A profusion of Bavli variants also exist. Even in the pericope you cited, Pesachim 103a, the form הילל can be found in manuscripts Vilna, Venice, Munchen 95, T-S F 1(1).55 and Hr. 15.37; but in the same section, the form הלל is found in manuscripts Munchen 6, JTS 1608 & 1623, Vatican 125 & 109, Columbia 294–295 and British Library Or. 5530.
In sum, Hillel can be found as both הלל or הילל. Plene and defective spellings were used interchangeably in many other cases as well, even in Tanakh, to a certain degree. Obviously, the issue of spelling the name was of little consequence to the scribes of these manuscripts and certainly the spellings do not affect the way the name was pronounced.