Jelly, 62 59 bytes

“ḢE{ɠs{Ƒ0Ṇṁỵ8ỊṢṂƊeLṫfIWẈḞ'ʠJ£ṗɱçoȧ?ƒnØẆƥṂ⁷ɱ’b12ĖŒṙị⁾ #s27Y

Try it online!

How it works

“ḢE{ɠs{Ƒ0Ṇṁỵ8ỊṢṂƊeLṫfIWẈḞ'ʠJ£ṗɱçoȧ?ƒnØẆƥṂ⁷ɱ’

is a bijective base-250 integer literal, which uses the characters in Jelly's code page as digits.

The literal encodes the integer $scriptsize 10250938842396786963034911279002199266186481794751691439873548591280332406943905758890346943197901909163$, which b12 converts to duodecimal, yielding $scriptsize 30b620b40ba54a64771433841333139413468423468423467433467433466b466b465453465363431424b43896860ba3_{12}$.

There are some zeroes, because 14 spaces (e.g.) are encoded as 3 spaces, 0 hashes, and 11 spaces. This maintains the base small (the largest run consists of 17 hashes), without adding any additional logic to the decoder.

Ė (enumerate) prefixes every base-12 digit by its 1-based index, then Œṙ performs run-length decoding.

Now, for each integer in the resulting array, ị⁾ # indexes into the string
consisting of a space and a hash. Indexing is 1-based and modular in Jelly, so odd integers get mapped to a space, even ones to a hash.

Finally, s27 splits the unformatted character array into chunks of length 27, which Y separates by linefeeds.

R, 198, 173, 163 157 bytes

write(matrix(rep(rep(c(' ','#'),,89),utf8ToInt("NFMOJEDJFDGGADCCHDACCCACIDACDFHDBCDFHDBCDFGDCCDFGDCCDFFKDFFKDFEDECDFECFCDCADBDKDCHIFHQJCR")-64),27),1,27,,'')

Try it online!

Also R must honor the great Stan

Explanation :

Applying a run-length encoding (rle function) on the characters of the string (excluding 'n') returns 89 repetitions of the alternating <space>,<hash>,<space>,<hash>,<space>,<hash>....

The 89 repetitions are all values in the range [1,18] hence, summing 64 we get the code points of the letters [A...R] obtaining the string : "NFMOJEDJFDGGADCCHDACCCACIDACDFHDBCDFHDBCDFGDCCDFGDCCDFFKDFFKDFEDECDFECFCDCADBDKDCHIFHQJCR"

Hence the code basically does the opposite operations :

v <- utf8ToInt("NFMOJEDJFDGGADCC...")-64  # convert the string back to [1...18] vector

e <- rep(rep(c(' ','#'),,89),v)           # perform the inverse of rle function

m <- matrix(e,27)                         # arrange the expanded vector in a matrix of 27 rows

write(m,1,27,,'')                         # write each row of the matrix to stdout

Canvas, 74 73 71 bytes

ｑｃ２－Ａｚ↓n⁴╫ｍ┬ｆｆ╷＼_↘Tｔ)%⁵6＞Ｙy7pＱ∔Ｉ％SIŗ＋T＾≤?↔↖¶８^`O‾+Ｏ│≤n≡j↶82„２┬｛ #＠］∑‾＋ｎ

Try it here!

///, 170 bytes

/+/////*/  +)/'$# +(/"$!+'/" +&/!#+%/"!

!+$/*!+"/**+!/###/""('*

(!!!' 

"&#"!!&"

$#'$& &*

 !""& ! $ 

!"'& !%""&$%""&$%)$%)$%(&#%(&#%'&'!%'!"$"!

 &$#")

$&#"'!!*

(!!&#"

"$

Try it online!

C (gcc), 244 243 bytes

#define z <<10|117441415

x={8064,2097088,8142832,31473596,58751886,30 z,60 z,60 z,120 z,120 z,255 z,255 z,480 z,448 z,63897630,33423612,2097136,1835008};b(_,i){i&&b(_/2,i-1),putchar(" #"[_%2]);}main(_){_<19&&b(x[_-1],27)&puts("")&main(_+1);}

Try it online!

-1 Thanks to Peter Cordes

Uses bit compression.

Explanation:

The goal is to print a set of numbers in binary using space and # as digits which represent the logo. A bit of bash magic converts the logo to binary masks:

echo "ibase=2;$(<code which echoes the logo [see my bash solution for example]> | tr ' #' 01)" | bc

This results in the binary 'numbers' being:

8064,2097088,8142832,31473596,58751886,117472135,117502855,117502855,117564295,117564295,117702535,117702535,117932935,117900167,63897630,33423612,2097136,1835008

There is an obvious pattern in the middle where every line contains ### ### ###

We can save some space by compressing that middle section based on saving that pattern and OR-ing against it. In addition, all of those lines merely add some stuff to the left of the middle section, so we make the z macro which takes ?????????????? and converts it into ###??????????????### ###. This involves bitshifting left by 10 and OR-ing with the binary of that pattern, which is 117441415.

Now we can more easily understand the code:

#define z <<10|117441415 // Define z to be the compression defined above

x={                    // Define x to be an array storing each line's number

8064,2097088,8142832,    // The first 5 lines are uncompressed

31473596,58751886,

30 z,60 z,60 z,120 z,    // The middle 9 lines are z-compressed

120 z,255 z,255 z,480 z,

448 z,

63897630,33423612,       // The last 4 lines are uncompressed

2097136,1835008};

b(_,i){                  // we also need a function to print n-bit binary numbers

i&&                      // i is the bit counter, we recurse until its zero

b(_/2,i-1),              // each recursive call halves the input and decrements i

putchar(" #"[_%2]);}     // this just prints the correct character

main(_){                 // this is the main function, called as ./? will have 1 in _ (argc)

_<19?                    // if _ is less than 19 then...

  b(x[_-1],27),          // print the binary expansion of x[_-1]

  puts(""),              // print a new line

  main(_+1)              // recurse with _++

:0;}

Bubblegum, 71 66 bytes

0000000: 95 91 b7 01 c0 30 08 c0 76 9e 48 fb ff c6 54 41  .....0..v.H...TA

0000010: ac cd 68 43 4a 71 59 c5 f1 12 0c 26 ca fd 11 8f  ..hCJqY....&....

0000020: 1c fc fa 15 7c 06 e6 27 55 71 72 23 b5 1b a9 dd  ....|..'Uqr#....

0000030: 0e 68 36 9c bf 69 a5 96 2b 47 41 9e 25 4c dc 89  .h6..i..+GA.%L..

0000040: fd 05                                            ..

Try it online!

JavaScript (ES6), 173 170 bytes

_=>`tc

du

9a9k

58fe38

36h83676

6j83!h85!h85!f87!f87!dm96

6dm96

6b8b!b6d696

3858n8

5gjc

dy

d6`.split`!`.join`696

6`.replace(/./g,c=>'# '[(k=parseInt(c,36))&1].repeat(k/2))

Try it online!

Node.js, 163 bytes

Provided that an array of strings is a valid output:

_=>Buffer(`.&-/*%$*&$''!$##($!###!#)$!#$&($"#$&($"#$&'$##$&'$##$&&+$&&+$&%$%#$&%#&#$#!$"$+$#()&(1*#2`).map(c=>s+='# '[x^=1].repeat(c-32),s=x='')&&s.match(/.{27}/g)

Try it online!

Charcoal, 71 bytes

Ｐ⁸↓Ｐ-³↙←⁴↓Ｐ³↙Ｇ→³↓⁹←³#↘Ｐ⁴↘Ｐ⁶↘‖ＯＭ←←¹⁷↘Ｆ⁹«Ｐ⁺³›ι³↑Ｐ⁴↗¿⁼ι²Ｂ⁸±²»↘→ＵＯ³¦¹⁴ＵＭＫＡ#

Try it online! Link is to verbose version of code. Explanation:

Ｐ⁸↓Ｐ-³↙←⁴↓Ｐ³↙Ｇ→³↓⁹←³#↘Ｐ⁴↘Ｐ⁶

Output half of the circle.

↘‖ＯＭ←←¹⁷↘

Reflect it and draw in the bottom.

Ｆ⁹«Ｐ⁺³›ι³↑Ｐ⁴↗¿⁼ι²Ｂ⁸±²»

Draw the left arm and crossbar of the A.

↘→ＵＯ³¦¹⁴

Draw the right arm of the A.

ＵＭＫＡ#

Change all the characters to #s.

C (gcc), 174 168 164 bytes

-6 bytes thanks to Dennis.

i,j=88;main(l){while(j--)for(i="CJQHFIHCDKDBDACDCFCEFDCEDEFDKFFDKFFDCCDGFDCCDGFDCBDHFDCBDHFDCADICACCCADHCCDAGGDFJDEJOMFN"[j]-64;i;)putchar(l++%28?--i,35-j%2*3:10);}

Try it online!

Bash, 192 176 bytes

dc<<<"16i2o81C0000 81FFFF0 9FE00FC BCF001E F070387 F078387 F03FF87 F03FF87 F01E387 F01E387 F00F387 F00F387 F007B87 B807B8E 9E03FBC 87C3FF0 81FFFC0 8001F80f"|tr 01 ' #'|cut -c2-

Try it online!

-16 thanks to manatwork

This is similar to my C answer, except it just uses a raw base-16 compression and passes it through bc, then uses tr to convert 1 to # and 0 to space. Each row has 1 appended to it and stripped off of it to maintain alignment.

Unfortunately dc is shorter than bc.

J, 130 128 bytes

echo' #'{~18 27$;(_243{.2#.inv 92x#._32+a.i.])&>'!TYPW.ajz i8hIhXl''3lOH8GvV.C2Z{r/=,G';'"a*2ZDxRplkh2tzRakz.?ZwVmeOT6L^lFB^eyT'

Try it online!

Initial solution

J, 164 bytes

echo' #'{~18 27$,#:849239965469633263905532594449192007713271791872263657753301928240007 12380965417202148347902847903517734495157419855048834759608223758433386496x

Try it online!

T-SQL, 341 338 bytes

DECLARE @ CHAR(2000)=REPLACE(REPLACE(REPLACE(

'PRINT SPACE(14*6&$6*15&$4*5$4*10&$2*4$7*7$1*4&$1*3$8*4$1*3$3*3&*3$9*4$1*3$4

 *3&*3$8*4$2*3$4*3&*3$8*4$2*3$4*3&*3$7*4$3*3$4*3&*3$7*4$3*3$4*3&*3$6*11$4*3&

 *3$6*11$4*3&*3$5*4$5*3$4*3&*3$5*3$6*3$4*3&$1*4$2*4$11*4&$2*8$9*6&$6*17&$6*3)'

,'*',')+REPLICATE(''#'','),'$',')+SPACE('),'&',')+(''

''')EXEC(@)

The first 4 line breaks are for readability only, the final line break is part of a string literal.

Similar to my Adam West tribute, I've manually encoded a long string, and made the following replacements:

• 
  *7 gets replaced by +REPLICATE('#',7)
  


• 
  $4 gets replaced by +SPACE(4)
  


• 
  & gets replaced by a line break inside quotes

This results in a massive SQL command string:

PRINT SPACE(14)+REPLICATE('#',6)+('

')+SPACE(6)+REPLICATE('#',15)+('

')+SPACE(4)+REPLICATE('#',5)+SPACE(4)+REPLICATE('#',10)+('

')+SPACE(2)+REPLICATE('#',4)+SPACE(7)+REPLICATE('#',7)+SPACE(1)+REPLICATE('#',4)+('

')+SPACE(1)+REPLICATE('#',3)+SPACE(8)+REPLICATE('#',4)+SPACE(1)+REPLICATE('#',3)+SPACE(3)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(9)+REPLICATE('#',4)+SPACE(1)+REPLICATE('#',3)+SPACE(4)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(8)+REPLICATE('#',4)+SPACE(2)+REPLICATE('#',3)+SPACE(4)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(8)+REPLICATE('#',4)+SPACE(2)+REPLICATE('#',3)+SPACE(4)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(7)+REPLICATE('#',4)+SPACE(3)+REPLICATE('#',3)+SPACE(4)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(7)+REPLICATE('#',4)+SPACE(3)+REPLICATE('#',3)+SPACE(4)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(6)+REPLICATE('#',11)+SPACE(4)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(6)+REPLICATE('#',11)+SPACE(4)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(5)+REPLICATE('#',4)+SPACE(5)+REPLICATE('#',3)+SPACE(4)+REPLICATE('#',3)+('

')+REPLICATE('#',3)+SPACE(5)+REPLICATE('#',3)+SPACE(6)+REPLICATE('#',3)+SPACE(4)+REPLICATE('#',3)+('

')+SPACE(1)+REPLICATE('#',4)+SPACE(2)+REPLICATE('#',4)+SPACE(11)+REPLICATE('#',4)+('

')+SPACE(2)+REPLICATE('#',8)+SPACE(9)+REPLICATE('#',6)+('

')+SPACE(6)+REPLICATE('#',17)+('

')+SPACE(6)+REPLICATE('#',3)

Which, when run, produces the necessary output.

Long, but still better than my best set-based solution (463 bytes):

SELECT SPACE(a)+REPLICATE('#',b)+SPACE(c)+REPLICATE('#',d)

      +SPACE(e)+REPLICATE('#',f)+SPACE(g)+REPLICATE('#',h)

FROM(VALUES(7,0,7,6,0,0,0,0),(6,8,0,7,0,0,0,0),(4,5,4,5,0,5,0,0),(2,4,7,7,1,4,0,0),

           (1,3,8,4,1,3,3,3),(0,3,9,4,1,3,4,3),(0,3,8,4,2,3,4,3),(0,3,8,4,2,3,4,3),

           (0,3,7,4,3,3,4,3),(0,3,7,4,3,3,4,3),(0,3,6,6,0,5,4,3),(0,3,6,6,0,5,4,3),

           (0,3,5,4,5,3,4,3),(0,3,5,3,6,3,4,3),(1,4,2,4,5,0,6,4),(2,8,9,6,0,0,0,0),

           (6,9,0,8,0,0,0,0),(6,3,0,0,0,0,0,0))t(a,b,c,d,e,f,g,h)

Python 2, 131 bytes

00000000: 2363 6f64 696e 673a 4c31 0a70 7269 6e74  #coding:L1.print

00000010: 2778 da95 9141 0e5c 3021 0803 ef7d 4593  'x...A.!...}E.

00000020: feff 8f9b 5d14 5c6e f1b0 f622 7422 2890  ....].n..."t"(.

00000030: 2e7d 5a09 dc4b 19cb bc53 84d1 4a6a 5960  .}Z..K...S..JjY`

00000040: 116e 3f42 c290 b3f7 c0bc 76cf 549d 6ed8  .n?B......v.T.n.

00000050: f8fa 5f26 0b0e 8c93 d5cb 35f6 b1e7 a939  .._&......5....9

00000060: 9e98 e769 47b9 87d6 cdf5 5c30 3030 32c0  ...iG.....002.

00000070: 4029 5c6e 272e 6465 636f 6465 2827 7a69  @)n'.decode('zi

00000080: 7027 29                                  p')

Try it online!

05AB1E, 101 88 64 bytes

„# •∍ΔÎë,½=bOÅ.âαUΔ'òõƶαÔγλ#xÆ]~”FbćÁ˜Ð”wнQ_wā©•12вεN>yи}˜èJ27ô»

-24 bytes by creating a port of @Dennis♦' Jelly answer.

Try it online.

Explanation:

„#                     # Push string "# "

   •∍ΔÎë,½=bOÅ.âαUΔ'òõƶαÔγλ#xÆ]~”FbćÁ˜Ð”wнQ_wā©•

                      '# Push compressed integer 10250938842396786963034911279002199266186481794751691439873548591280332406943905758890346943197901909163

     12в               # Convert to Base-12 as list: [3,0,11,6,2,0,11,4,0,11,10,5,4,10,6,4,7,7,1,4,3,3,8,4,1,3,3,3,1,3,9,4,1,3,4,6,8,4,2,3,4,6,8,4,2,3,4,6,7,4,3,3,4,6,7,4,3,3,4,6,6,11,4,6,6,11,4,6,5,4,5,3,4,6,5,3,6,3,4,3,1,4,2,4,11,4,3,8,9,6,8,6,0,11,10,3]

        ε    }         # Map each `y` to:

         N>            #  The index+1

           yи          #  Repeated `y` amount of times

              ˜        # Flatten the list

               è       # Index each in the string "# " (with automatic wraparound)

                J      # Join everything together

                 27ô   # Split into parts of length 27

                    »  # And join by newlines

See this 05AB1E tip of mine (sections How to compress large integers? and How to compress integer lists?) to understand how the compression of the integer and list works.

Haskell, 170 163 bytes

Edit: -7 bytes thanks to @Ørjan Johansen

m=<<"Mc Ul WeWg YdTgZa ZcSdZcX cRdZcW cSdYcW cSdYcW cTdXcW cTdXcW cUkW cUkW cVdVcW cVcUcW ZdYdPa YhRc Un U "

m ' '="###n"

m c=(' '<$['Z','Y'..c])++('#'<$['a'..c])

Try it online!

Spaces are encoded as uppercase characters (length: Z down to char), hash signs as lowercase characters (length: a to char) and the last three # of each line plus newline as a space. Function m decodes it.

Perl 5, 181 bytes

say+(sprintf"%28b",oct"0x$_")=~y/10/# /r for"0003f0003fff800f87fe03c07f78700f71ce00f70ee01e70ee01e70ee03c70ee03c70ee07ff0ee07ff0ee0f070ee0e070e79e003c3fc01f803fffe00380000"=~/.{7}/g

Try it online!

Perl 6, 136 116 bytes

:100['>	b3V$^%&!U3X/_9-D>Z(+, OD_=D,-:&3S>#OL

8C"> c H'.ords].base(2)~~TR/1/ /.comb(27)>>.say

Try it online!

Parses a base 100 number from the ordinal values of the string, converts to base 2 and replaces the 1s with spaces. This uses zeroes as the main character.

Explanation:

:100['...'.ords]   # Convert the bytes of the string to base 100

.base(2)           # Convert to base 2

~~TR/1/ /          # Translate 1s to spaces

                   # Alternatively, this could be .split(1)

.comb(27)          # Split to strings of length 27

>>.say             # Print each on a newline

C# (.NET Core), 199 bytes

_=>{var r="";for(int i=0,j,k=0;i<88;i++)for(j=0;j++<"0(/1,'&,(&))#&%%*&#%%%#%+&#%&(*&$%&(*&$%&()&%%&()&%%&((-&((-&('&'%&('%(%&%#&$&-&%*+(*3,%"[i]-34;){r+=i%2<1?' ':'#';if(++k%27<1)r+='n';}return r;}

Try it online!

Uses the same approach as my solution to the tribute to Adam West.

PHP, 286 bytes

<?=preg_replace_callback('~(d*)(D)~',function($m){return str_repeat($m[2],$m[1]?:1);},str_replace(["

",'~','^',';'],["^

",'^4 ','3#','4#'],"14 ^

6 12#

4 5#4 7#

2 ; 6 7# #

 ^8 ; ^3 

^9 ; ~

^8 ;2 ~

^8 ;2 ~

^7 ;3 ~

^7 ;3 ~

^6 11#4 

^6 11#4 

^5 ;5 ~

^5 ^6 ~

 ;2 ;11 #

2 8#9 ^

6 1;

6 ^"));

Try it online!

MATLAB : 144 Bytes

reshape(repelem([repmat(' #',1,44),' '],'NFMOJEDJFDGGADCCHDACCCACIDACDFHDBCDFHDBCDFGDCCDFGDCCDFFKDFFKDFEDECDFECFCDCADBDKDCHIFHQJCR'-64),27,)'

Try it online! (Technically in Octave)

Explanation:

This uses the same strategy as digEmAll in R, just with MATLAB syntax. The main difference is that MATLAB have automatic conversion from characters to integers.

deflate, 79 bytes

eJyVkcsJADAIQ++ZIpD9dyxUqrGUgjlpHvglXdqCJymk1yEiaRXEIOXzuBHCiKObReUxUzYaMdt2wmTBg/FmNXndgLRbNvL7ifsLfMw6iQ==