Berk (@_berkt) header
Berk (@_berkt) profile picture

bsc @metu_odtu ceng | msc @usfca cs | ai @stanford interested in graph-based behavioral targeting & recommendation systems

Menlo Park, CA
Joined December 2020
Photo by Berk (@_berkt)Photo by Berk (@_berkt)Photo by Berk (@_berkt)Photo by Berk (@_berkt)Photo by Berk (@_berkt)Photo by Berk (@_berkt)Photo by Berk (@_berkt)Photo by Berk (@_berkt)Photo by Berk (@_berkt)
Got the official letter 🥳🎉
Life update: I'll be working as a teaching assistant at Stanford University next quarter to help students with their classes! Very excited.
1
79
3,661
Life update: I'll be working as a teaching assistant at Stanford University next quarter to help students with their classes! Very excited.
3
1
61
5,606
Officially signed a new offer. I’ll be commuting to Menlo Park two days a week from now on
2
18
1,439
Okulda her gun arabayi parkettigim yere geldim park yeri bulma sorununu birebir yapmislar
a Waymo run down San Francisco's iconic Lombard Street. built with fable 5.5 + threejs play here: teleoperator.mindblown.ai
13
1,634
Stanford CS229 machine learning dersinden B+ almanin verdigi yetkiye dayanarak bugun gidip envai cesit stanford yazili tisort, sweatshirt, kahve kupasi, anahtarlik vs aldim. Her gun yanima birini alarak ciksam yeterli olur diye dusunuyorum
1
26
894
I'm excited to announce that following works have been accepted to ECCV 2026. I will be attending the conference in person and would love to meet up with fellow researchers! 1) 𝐋𝐚𝐲𝐞𝐫-𝐀𝐰𝐚𝐫𝐞 𝐕𝐢𝐝𝐞𝐨 𝐂𝐨𝐦𝐩𝐨𝐬𝐢𝐭𝐢𝐨𝐧 𝐯𝐢𝐚 𝐒𝐩𝐥𝐢𝐭-𝐭𝐡𝐞𝐧-𝐌𝐞𝐫𝐠𝐞 This approach enables the composition of 𝒂𝒏𝒚 𝒇𝒐𝒓𝒆𝒈𝒓𝒐𝒖𝒏𝒅 𝒗𝒊𝒅𝒆𝒐 with 𝒂𝒏𝒚 𝒃𝒂𝒄𝒌𝒈𝒓𝒐𝒖𝒏𝒅 𝒗𝒊𝒅𝒆𝒐. We revisit the problem of video composition through the lens of GenAI. Unlike conventional methods that rely on annotated datasets or handcrafted rules, Split-then-Merge (StM) splits a large unlabeled corpus of videos into dynamic foreground and background layers, then merges them to learn how dynamic subjects interact with diverse scenes. 🌐 Project webpage: split-then-merge.github.io/ 📜 arXiv: arxiv.org/abs/2511.20809 2) 𝐍𝐚𝐫𝐫𝐚𝐭𝐢𝐯𝐞-𝐃𝐫𝐢𝐯𝐞𝐧 𝐏𝐚𝐩𝐞𝐫-𝐭𝐨-𝐒𝐥𝐢𝐝𝐞 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐯𝐢𝐚 𝐀𝐫𝐜𝐃𝐞𝐜𝐤 𝐖𝐢𝐭𝐡 𝐀𝐫𝐜𝐃𝐞𝐜𝐤, 𝐲𝐨𝐮 𝐜𝐚𝐧 𝐭𝐮𝐫𝐧 𝐚𝐧𝐲 𝐩𝐚𝐩𝐞𝐫 𝐢𝐧𝐭𝐨 𝐚 𝐩𝐫𝐞𝐬𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧 𝐰𝐢𝐭𝐡 𝐚 𝐜𝐨𝐦𝐩𝐞𝐥𝐥𝐢𝐧𝐠 𝐧𝐚𝐫𝐫𝐚𝐭𝐢𝐯𝐞. By combining the paper’s narrative and assets with engaging visual layouts, ArcDeck generates appealing slide decks tailored to the chosen presentation duration and target audience. 🌐 Project webpage: arcdeck.org/ 💻 GitHub & Claude Skill: github.com/RehgLab/ArcDeck 🤗 HuggingFace Dataset: huggingface.co/datasets/ArcD… 📜 arXiv: arxiv.org/abs/2604.11969 #ECCV2026 @eccvconf
2
1
34
1,355
kaliteli bir open back kulaklik alayim dedim amplifier gerekiyormus kullanmak icin. bir de onu siparis ettik bakalim studyo kurmaya kadar gitmez bu is umarim
1
816
Life update: az evvel green cardimizi almis bulunmaktayiz. Siradaki adim vatandaslik
22
1,123
Dugune ozel supra konvoyu
2
18
986
Hava guzel
18
623
Tokyo drift han reisin șahsi arabasıyla hatıra fotosu
9
649
Sonoma gt world challenge’da display car olarak yerimizi aldik
10
14,196
Alandayiz
5
628
Bir takim araba isleri ve barbeku
12
491
Full downforce wingimizin montajini tamamladik
14
3,153
New year new color
23
581
Sonbahar bitmeden ufak bir foto cekme denemesi
16
2,096
Bugunku konumumuz
1
316
O sirada bizim underground supra meetingin ortama giris sekli
Bir içerik üreticisi, katıldığı Porsche Club İstanbul etkinliğini paylaştı.
10
1,401