Cloud video rendering service on GPU with a Telegram bot, local Whisper, and API integrations
Cloud video rendering service on GPU with a Telegram bot, local Whisper, and integrations with Fal AI (Flux Schnell and Wan 2)
1. Project Goal
Create a cloud video rendering service with a Telegram interface that:
Accepts files and instructions via a Telegram bot
Fetches images via the Fal AI Flux Schnell API
Fetches videos via the Fal AI API (Wan 2 model)
Locally generates subtitles (Whisper)
Renders everything on GPU (via ffmpeg/OpenCV/SDK)
Delivers the final result via Google Drive
2. Technology Stack
Telegram Bot API (Python: [aiogram, pyrogram] or Node.js)
ffmpeg with GPU support (NVENC, CUDA, VideoToolbox)
OpenCV with CUDA (if needed for custom frame processing)
Fastvideo SDK (if acceleration is needed, budget permitting)
OpenAI Whisper (locally, CPU/GPU, via Python binding)
Fal AI API (Flux Schnell — images, Wan 2 — videos)
Google Drive API (uploading finished videos)
Cloud: DigitalOcean (or AWS/GCP), must have GPU instances
Docker (containers for isolating render workers)
(Optional) Kubernetes for automatic scaling
3. Main Logic of Operation
3.1. Telegram Bot
Receives instructions, user files (text, audio, video, parameters).
Displays statuses (“queue”, “rendering”, “ready”).
Allows selection of template, style, and setting text for subtitles.
3.2. Render Pipeline
User request received → added to queue (queue on Redis/DB).
Images fetched via Fal AI Flux Schnell API
Send prompt, receive url/jpg/png.
Video fetched via Fal AI Wan 2
API request, receive mp4/mov or url.
Subtitles generated via Whisper
Audio (if needed) is fed into Whisper (locally on GPU/CPU).
Result — srt/txt file.
Video rendering on GPU
All elements are assembled (video, images, audio, subtitles) into one video clip.
ffmpeg with full GPU support.
Optional custom processing via OpenCV CUDA.
Final file uploaded to Google Drive
A unique folder/link is created.
User is sent a link in Telegram
4. Technical Requirements for Rendering
Rendering 10-15 videos (1 hour each) in parallel
ffmpeg runs with NVENC/CUDA parameters
Adding audio/video/subtitles — only via GPU
Whisper installed locally (via Docker or system-wide)
Fal AI API — integration via REST, error handling/retry
All intermediate files are deleted after rendering is complete
5. Architecture (schematic)
User (Telegram)
Telegram Bot
↕Render Queue Manager (render queue, status monitoring)
↕Render Worker (Docker):
Fetches data from Fal AI (images/videos)
Calls Whisper for subtitles
Merges video (ffmpeg on GPU)
Uploads result to Google Drive
Notifies Queue Manager/bot
6. Minimum Server Requirements
DigitalOcean GPU droplet (8+ CPU cores, 32+ GB RAM, 1+ GPU, NVMe SSD)
Docker installed
Python 3.10+, Node.js (if needed for bots)
ffmpeg with CUDA/NVENC support
OpenCV with CUDA
Whisper (installation via pip + models)
Internet access for Fal AI and Google Drive APIs
7. Sample Pipeline (pseudocode)
# 1. Fetch image via Fal AI Flux Schnell
image_url = fal_api.get_image(prompt)
download(image_url)
# 2. Fetch video via Fal AI Wan 2
video_url = fal_api.get_video(params)
download(video_url)
# 3. Whisper (locally) -> get subtitles
subtitles = whisper.transcribe(audio_file)
# 4. ffmpeg (GPU) — assemble video
cmd = f"ffmpeg -hwaccel cuda -i video.mp4 -i image.jpg -vf 'subtitles=subs.srt' -c:v h264_nvenc output.mp4"
run(cmd)
# 5. Upload to Google Drive, get link
link = google_drive.upload('output.mp4')
telegram_bot.send_message(user_id, link)
8. What the Developer Needs to Do
Set up the Telegram bot (or refine it if it already exists)
Write a handler for integration with Fal AI (2 endpoints)
Prepare a render worker (Docker) that:
Fetches media from the API
Calls local Whisper
Generates subtitles in SRT
Runs ffmpeg on GPU with all parameters
Uploads video to Google Drive
Sends statuses back to Telegram
Ensure scalability and parallelism (multiple Docker workers)
Documentation for setup and deployment
9. Documentation/Reports
Describe deployment on DigitalOcean (or another cloud), instructions for raising containers, connecting to APIs
README with command examples and testing
10. Deadlines and Budget
Propose deadlines for MVP implementation (basic flow, 1-2 workers)
Estimate budget for launching on DigitalOcean for 1 month of operation (guideline: 1-2 GPU droplets)
Ready to provide additional details (Fal AI API, prompt examples, video/image examples for testing) — upon request.
-
12 days334 USD
1017 2 1 12 days334 USD
Hello! 👋
I have practical experience in data parsing (real projects, working with APIs, structuring and processing information) and creating Telegram bots. I have also done several PET projects using AI:
BizzAi — a bot for generating business responses;
fuckupcoach — an interactive coaching bot.
I know how to work with API integrations, queues, and automation, and I am ready to implement an MVP for your service:
…
A Telegram bot for receiving data and providing results;
Integration with Fal AI API;
Local work with Whisper for generating subtitles;
Organization of a GPU render pipeline (ffmpeg, OpenCV, Docker);
Uploading finished results to Google Drive and notifying the user.
I can quickly get to work, master specific tools (CUDA, ffmpeg with GPU, scaling workers), and document the process.
I would be happy to join the implementation of this project 🚀
-
1 day601 USD
1595 7 0 1 day601 USD
Good day!
My name is Roman, and I am among the top 5 developers in the "Artificial Intelligence and Machine Learning" category out of ~1600 specialists on the platform.
I guarantee:
- Fast and quality task execution
- Strict adherence to deadlines
- Regular communication throughout the entire process
I would be happy to discuss the details of your project in private messages.
-
14 days312 USD
330 2 0 14 days312 USD
Hello.
I have experience in implementing AI conversations - I can show dialogues of ready-made chats and results.
I also have developments of maximally realistic photographs - I am also ready to show the results.
I think from the description we understand each other about what your service is for - and we will quickly find a common language.
-
1 day45 USD
10063 117 0 1 day45 USD
Hello.
I am developing bots for Telegram in NodeJS. I am ready to take it on. Write to me, we will discuss.
-
7 days312 USD
954 5 1 7 days312 USD
Good afternoon, I am ready to take on the task, I have a similar ready bot but for different tasks, but I can work with the Fal API bot which is connected to Halioai. Here is the link to my portfolio https://freelancehunt.com/showcase/work/bot-dlya-avtomatichnoyi-generatsiyi-video-po/1973705.html
I am ready to redo it for your tasks, the API for testing will definitely be needed, the costs for tests will be approximately $10 and above.
Access to the server, if you don't have it, please register and give me access, I will set up the environment myself, in general, with the network that you described in the technical specifications, I will be able to assemble it.
My 1 day costs 2000 UAH.
I will most likely finish earlier than 7 days, but I need a few days for testing to make sure everything is good.
Link to the work video
Google Drive
Current freelance projects in the category Bot Development
Automation setup: Form on the website → WhatsApp bot → CRM (Make + WATI)Project Description: There is a transportation company website. Clients submit requests through a form on the site — the form is sent to email. We need to set up the following automation:Stage 1 — Form → WhatsApp — When a new request comes to the email from the website, the… Web Programming, Bot Development ∙ 15 hours 46 minutes back ∙ 29 proposals |
Telegram chatbot for a closed club
22 USD
Good day! I need to set up a bot for Telegram. Task: selling access to a closed channel. The bot should accept payments and after successful payment, issue a one-time invitation link. It is also necessary to set up automatic removal of a participant from the channel when their… Bot Development ∙ 3 days 2 hours back ∙ 93 proposals |
Mini apps Telegram. StoreGood afternoon An application is required in Telegram (online store) Example - essentially 1 to 1 like here brandspacebot Your experience is important, the stack you propose to use and why Price and timeframe approximately to start, I will be waiting for questions Web Programming, Bot Development ∙ 3 days 4 hours back ∙ 89 proposals |
Bots"I am looking for a programmer to create a turnkey Telegram bot. The bot should be an intelligent conversationalist based on AI (Gemini or ChatGPT), helping people on any topic and in any language. Payment (Monobank or LiqPay) for subscription needs to be integrated as well. I… Bot Development ∙ 3 days 13 hours back ∙ 61 proposals |
Need an AI agent for cold calls in Ukrainian.
445 USD
Need an AI agent for cold calls in Ukrainian. The topic is biofertilizers. The agent should call clients, work with a script, answer typical questions, and transfer interested clients to the manager. Need the cost of setup and monthly expenses. Bot Development ∙ 4 days 1 hour back ∙ 37 proposals |
Ukraine