Bot+TikTok
Project Description:
It is necessary to create a minimal prototype of a service in the style of “Tinder for humor,” which will find people with similar comedic tastes based on likes in TikTok. All through Telegram (bot).
⸻
✅ What needs to be done:
1. Telegram bot:
• Receiving TikTok nickname
• User instructions on how to view likes
• Notification when the profile is ready
• Simple match output based on humor
2. TikTok scraper:
• Retrieve public likes of a user (for example, 50–100 items)
• Using Playwright or another headless browser
• Collect: description, hashtags, author, ID, sound, and if possible — covers
3. AI humor analysis:
• Use SentenceTransformers and CLIP for content vectorization
• Store vectors in memory or FAISS
• Find matches between users (most similar vectors)
4. Matching:
• If two users have high similarity — send them a match
• No full chat yet — just display information about the other user
5. Admin panel / CLI:
• Display:
• Number of users
• Number of profiles processed / in queue
• Gender statistics
• Processing queue
• List of matches
• GPU load (if used)
• Can be via CLI (Typer / Textual) — not necessarily web-based
⸻
💻 Technologies (stack):
• Python 3.10+
• FastAPI (or aiohttp)
• aiogram / pyrogram
• Playwright
• SentenceTransformers + CLIP + Faiss
• SQLite (possibly PostgreSQL later)
• Admin panel via CLI (Textual or Typer)
Format:
• Development on a local server (Ryzen 9 + 3080 Ti / 4080)
• No cloud, no deployment to production
• Just to test the idea
-
9 days349 USD
536 3 2 9 days349 USDGood afternoon!
Thank you for the detailed description — the idea of "Tinder with humor" is very interesting. To plan the work accurately, please clarify:
Expected number of users at launch (for example, 50–100 or several thousand immediately)?
How often should likes from TikTok be updated (once at registration or periodically)?
Are there any requirements for the bot's response time — the maximum allowable delay between the request and the match output?
…
I will be happy to refine the proposal based on your feedback!
-
7 days639 USD
1616 8 0 7 days639 USDHello
I am a developer in the field of ML/DL & Web Dev | Bot Dev | Web Scraping. Ready to complete your project. Write to me, let's discuss.
-
10 days349 USD
593 15 0 10 days349 USDHello.
I will not go into details of the technological stack, I will only say that I placed a bet on the previous project and since the case was interesting, I already implemented a parser for testing.
From my side, I suggest an interesting architectural solution for the MVP model where you have a separate bot that goes to the parser, and it only communicates with the user and searches for a match based on the ready database.
0) Adding a user and checking if likes are open...
1) The user registers and starts analyzing their profile, we defer parsing and insert data into the database. Initially, this is just a TikTok module, but other modules can be connected, and different analysis models for videos or sounds can be integrated.
2) We have a bot that will primarily suggest more relevant matches as long as there are matches in the database and the similarity index is not below a set threshold, say 0.6.
3) Depending on whether the user liked or not, we can share another user's profile (like in Dvanchik).
Advantages include a local PC with a similar 4070Ti + i9 11900, so I can immediately give feedback on processing. However, the first user processing with their TikTok may take up to 10-30 minutes.
…
P.S. While implementing locally is the best choice for MVP, there is one caveat: you will still need to connect some proxies for TikTok parsing, otherwise the IP will quickly get banned during the pilot launch.
Regarding the humorous aspect of the cover or sound, it is a very subjective decision, and even AI is a large R&D area specific to your project, where you will need to experiment afterwards.
-
1 day29 USD
414 1 0 1 day29 USDHello, I am ready to do it. I can also offer development in a more efficient programming language. I suggest discussing all the details in private messages.
-
7 days29 USD
198 7 days29 USDHello. I am ready to undertake the development of the MVP for the "Tinder for Humor" service. I will implement a Telegram bot with TikTok nickname processing and instructions, collect likes using Playwright, vectorize content with SentenceTransformers and CLIP, store and search for similarity using Faiss, provide simple match results, as well as a CLI admin panel using Typer or Textual. The work will be on a local server, without clouds or deployment. I am ready to discuss details and deadlines.
-
10 days349 USD
173 10 days349 USDHello, Nice to meet you.
I have a enough experience in python, telegam bot development and can build your task successfully.
Let's discuss in more detail
Best Regards
-
10 days319 USD
1520 2 0 10 days319 USDHello! The idea of the project sounds really unusual and interesting — to create a "Tinder for humor" based on likes from TikTok, with a Telegram bot, analysis through neural networks, and selection based on vector similarity. I am ready to take on the implementation of the MVP: bot (aiogram), scraper (Playwright), vectorization (SentenceTransformers + CLIP), similarity database (FAISS), simple CLI admin panel (Typer). Everything locally, without clouds — perfect for your task.
But I want to honestly point out one important aspect that, in my opinion, requires special attention:
A like on TikTok is not structured text, but a video. Therefore, the extracted textual context for humor analysis is limited.
Yes, we can collect descriptions (captions), hashtags, the author, and (if possible) thumbnails. This provides some context that can be vectorized and compared. But how accurately this set of metadata will reflect the user's humorous preferences — is still an experimental question. I would note this as R&D risk: it can be tried, but the results are better evaluated along the way.
Otherwise — the task is clear to me, the technologies are familiar (there is a similar side project with NLP only without scraping via API), and the tools are suitable. Time estimate: about 10–12 working days (2–2.5 weeks) for a full working prototype, without production deployment.
…
Regarding cost:
Minimum (if the hypothesis doesn't work / we stick to basic functionality): $250
Full MVP with decent accuracy and matching logic: $800–1000
If it’s important for you to quickly test the hypothesis and see how it works — I am ready to get involved.
Current freelance projects in the category Data Parsing
Parsing products from supermarketsNeed a specialist (parsing + Google Sheets + automation). Goal: Create a system for automatic retrieval and updating of food prices from Silpo and NOVUS supermarkets with data output to Google Sheets. What needs to be implemented: Create a main Google Sheets table with a product… Data Parsing, Information Gathering ∙ 59 minutes back ∙ 19 proposals |
LinkedIn mailing
120 USD
LinkedIn Mailing It is necessary to launch a LinkedIn mailing to the specified database. 30,000+ contacts We are waiting for proposals Email Marketing, Data Parsing ∙ 1 day 8 hours back ∙ 9 proposals |
Parsing company emails from a specific industry
113 USD
Find contacts (e-mail, phone number) of companies engaged in the specified business mentioned in this Google document https://docs.google.com/document/d/1_d-iLxLSL3gjRvo1GmUpWNm0Pm5YoKGcEvaiDc_2I0g/edit?usp=sharing The result should be in the form of a table with the following… Data Parsing ∙ 2 days 8 hours back ∙ 31 proposals |
Need a parser for the online store https://www.lcsc.com/It is necessary to regularly (once a month, or upon script launch) obtain up-to-date information about the products available in the store. https://www.lcsc.com/ from the catalog of all sections.… Data Parsing ∙ 3 days 14 hours back ∙ 43 proposals |
OpenCart — rental catalog of special equipment
135 USD
OpenCart — Equipment Rental Catalog Need to launch an equipment rental catalog on OpenCart. Theme: excavators cherry pickers forklifts generators cranes scaffolding other construction equipment. It is preferable that you already have a ready-made template or developments… Web Programming, Data Parsing ∙ 4 days 6 hours back ∙ 59 proposals |
Pozsony