Ahmed Mansour
@sourmans.bsky.social
Building Hyper-personalized video outreach for head hunters demo here https://loom.li pitch here https://share.descript.com/view/PxMCAgGSRJt
Pinned
Ahmed Mansour
@sourmans.bsky.social
· Nov 24
personalized candidate videos
loomli books more candidates by generating videos of you speaking to each candidate's LinkedIn profile
loom.li
Hi Friends 👋
building video gen ai loom.li for recruiters
graduated 2014 McMaster - Mechatronics Eng & Entrepreneurship
7 years @ NXP Semiconductors
Linux Kernel dev Networking HW
ML dev for ASIC computer vision in cars
Fulltime founder since 2021
(more 👇)
building video gen ai loom.li for recruiters
graduated 2014 McMaster - Mechatronics Eng & Entrepreneurship
7 years @ NXP Semiconductors
Linux Kernel dev Networking HW
ML dev for ASIC computer vision in cars
Fulltime founder since 2021
(more 👇)
took me a while to figure out how to share auth status across my subdomains using supabase authentication. I set up this repo that creates an env where you can test across subdomain auth very quickly locally to make sure it works in prod. Enjoy
demo: youtu.be/AhMpIxzF0Yk
github.com/sourman/exam...
demo: youtu.be/AhMpIxzF0Yk
github.com/sourman/exam...
GitHub - sourman/example-supabase-auth-across-subdomains: Example showing how to setup supabase auth so it propagates automatically across subdomains
Example showing how to setup supabase auth so it propagates automatically across subdomains - sourman/example-supabase-auth-across-subdomains
github.com
February 11, 2025 at 3:36 AM
took me a while to figure out how to share auth status across my subdomains using supabase authentication. I set up this repo that creates an env where you can test across subdomain auth very quickly locally to make sure it works in prod. Enjoy
demo: youtu.be/AhMpIxzF0Yk
github.com/sourman/exam...
demo: youtu.be/AhMpIxzF0Yk
github.com/sourman/exam...
- writing python using LLMS is OP
- writing React code using LLMs is like working with a lobotomized Senior Software Engineer. Knows the big words and rams them into concoctions of hairy painful bugs xD
I think the key is that React does a lot of implicit work which LLMs struggle with
- writing React code using LLMs is like working with a lobotomized Senior Software Engineer. Knows the big words and rams them into concoctions of hairy painful bugs xD
I think the key is that React does a lot of implicit work which LLMs struggle with
January 26, 2025 at 6:13 AM
- writing python using LLMS is OP
- writing React code using LLMs is like working with a lobotomized Senior Software Engineer. Knows the big words and rams them into concoctions of hairy painful bugs xD
I think the key is that React does a lot of implicit work which LLMs struggle with
- writing React code using LLMs is like working with a lobotomized Senior Software Engineer. Knows the big words and rams them into concoctions of hairy painful bugs xD
I think the key is that React does a lot of implicit work which LLMs struggle with
Tip: 2x improvement for debugging using Cursor Composer
Main idea: Run your server in `Terminal in Editor Area`
No Composer `agent` and `normal` have access to live debug prints
Main idea: Run your server in `Terminal in Editor Area`
No Composer `agent` and `normal` have access to live debug prints
January 25, 2025 at 1:03 AM
Tip: 2x improvement for debugging using Cursor Composer
Main idea: Run your server in `Terminal in Editor Area`
No Composer `agent` and `normal` have access to live debug prints
Main idea: Run your server in `Terminal in Editor Area`
No Composer `agent` and `normal` have access to live debug prints
Thanks to Aiden of million.dev for keeping me humble
January 23, 2025 at 6:41 AM
Thanks to Aiden of million.dev for keeping me humble
Google cloud gave ma a generous 2000 USD in credits for my startup so I gave the docker running solutions a try
1. `Cloud Run` Service <- good for persistent micro services
2. `Cloud Run` Jobs <- auto scale, 100% clean slate between runs
3. GKE (Google Kubernetes Engine) <- You admin the cluster
1. `Cloud Run` Service <- good for persistent micro services
2. `Cloud Run` Jobs <- auto scale, 100% clean slate between runs
3. GKE (Google Kubernetes Engine) <- You admin the cluster
January 19, 2025 at 7:22 PM
Google cloud gave ma a generous 2000 USD in credits for my startup so I gave the docker running solutions a try
1. `Cloud Run` Service <- good for persistent micro services
2. `Cloud Run` Jobs <- auto scale, 100% clean slate between runs
3. GKE (Google Kubernetes Engine) <- You admin the cluster
1. `Cloud Run` Service <- good for persistent micro services
2. `Cloud Run` Jobs <- auto scale, 100% clean slate between runs
3. GKE (Google Kubernetes Engine) <- You admin the cluster
Just realized gmail auto switches the spell checker from english to other languages automatically 🤯
This was not the case before! pretty cool
My French is not great but I manage :P
This was not the case before! pretty cool
My French is not great but I manage :P
January 18, 2025 at 4:39 AM
Just realized gmail auto switches the spell checker from english to other languages automatically 🤯
This was not the case before! pretty cool
My French is not great but I manage :P
This was not the case before! pretty cool
My French is not great but I manage :P
new python is colorful :) I like this new prompt. Also it finally accepts quit without the ()
January 13, 2025 at 3:24 PM
new python is colorful :) I like this new prompt. Also it finally accepts quit without the ()
Getting close. Now connecting the Text-To-Speach APIs. Interestingly I found elevenlabs insta voice clones to be not good enough for my use case. gave me a most british accent
January 7, 2025 at 10:45 PM
Getting close. Now connecting the Text-To-Speach APIs. Interestingly I found elevenlabs insta voice clones to be not good enough for my use case. gave me a most british accent
python people have the most fun with library names :)
imagine building a library called fuzzywuzzy. What an excellent mood
imagine building a library called fuzzywuzzy. What an excellent mood
January 7, 2025 at 10:43 PM
python people have the most fun with library names :)
imagine building a library called fuzzywuzzy. What an excellent mood
imagine building a library called fuzzywuzzy. What an excellent mood
I was curious about the activity level on bsky. There was an obvious post election increase and an expected drop, but currently it is stable at 2.5 million Daily Active Users. Very good. That is an additional 0.6 million DAU gained from the election fallout. Source bsky.jazco.dev/stats
January 6, 2025 at 6:00 PM
I was curious about the activity level on bsky. There was an obvious post election increase and an expected drop, but currently it is stable at 2.5 million Daily Active Users. Very good. That is an additional 0.6 million DAU gained from the election fallout. Source bsky.jazco.dev/stats
man, python array structs like this one are simply beautiful and succinct 🥹
January 6, 2025 at 1:40 AM
man, python array structs like this one are simply beautiful and succinct 🥹
Best app I installed on my phone
Takes wildcards and blocks robo calls automatically.
I get 5 of these a day and usually during my most focused productive hours
App is called YACB yet another call blocker
Takes wildcards and blocks robo calls automatically.
I get 5 of these a day and usually during my most focused productive hours
App is called YACB yet another call blocker
January 3, 2025 at 8:53 PM
Best app I installed on my phone
Takes wildcards and blocks robo calls automatically.
I get 5 of these a day and usually during my most focused productive hours
App is called YACB yet another call blocker
Takes wildcards and blocks robo calls automatically.
I get 5 of these a day and usually during my most focused productive hours
App is called YACB yet another call blocker
cursor sees your staged changes and writes decent commit messages
January 1, 2025 at 8:39 PM
cursor sees your staged changes and writes decent commit messages
looking much more human and correctly navigating the dreaded LI HTML soup
January 1, 2025 at 5:37 PM
looking much more human and correctly navigating the dreaded LI HTML soup
Automatic navigation + Scraping + scrolling + highlighting all working now. Just needs a bit of a human touch
January 1, 2025 at 2:03 PM
Automatic navigation + Scraping + scrolling + highlighting all working now. Just needs a bit of a human touch
Just figured out what's taking 80 GB of my disk space. Docker for Windows is an afterthought project and it hides junk inside of the AppData folder on Windows. For some reason Windows does not correctly attribute this disk usage to docker if you use the built-in disk utilization viewer
January 1, 2025 at 4:12 AM
Just figured out what's taking 80 GB of my disk space. Docker for Windows is an afterthought project and it hides junk inside of the AppData folder on Windows. For some reason Windows does not correctly attribute this disk usage to docker if you use the built-in disk utilization viewer
Asked o1-mini to look at my codebase and create a mermaid diagram of how functions call one another then pasted the text into stacked.io to render the diagram below. useful for finding violation of "separation of concerns" and orphaned functions
December 29, 2024 at 1:58 PM
Asked o1-mini to look at my codebase and create a mermaid diagram of how functions call one another then pasted the text into stacked.io to render the diagram below. useful for finding violation of "separation of concerns" and orphaned functions
Reposted by Ahmed Mansour
When is this monstrosity going to end?
December 27, 2024 at 11:55 PM
When is this monstrosity going to end?
gemini models just added to cursor. anyone play with them for coding tasks? any feedback?
December 25, 2024 at 11:05 AM
gemini models just added to cursor. anyone play with them for coding tasks? any feedback?
Reposted by Ahmed Mansour
This week we released ModernBERT, the first encoder to reach SOTA on most common benchmarks across language understanding, retrieval, and code, while running twice as fast as DeBERTaV3 on short context and three times faster than NomicBERT & GTE on long context.
December 22, 2024 at 6:12 AM
This week we released ModernBERT, the first encoder to reach SOTA on most common benchmarks across language understanding, retrieval, and code, while running twice as fast as DeBERTaV3 on short context and three times faster than NomicBERT & GTE on long context.
Fresh Cursor release **Yolo mode**. What could possibly go wrong xD Will post updates
December 20, 2024 at 5:00 PM
Fresh Cursor release **Yolo mode**. What could possibly go wrong xD Will post updates
The 3B Qwen LLM is honestly sweetspot performance vs compute time. Running it in a Docker container
December 19, 2024 at 3:25 AM
The 3B Qwen LLM is honestly sweetspot performance vs compute time. Running it in a Docker container
Giving local LLM a spin
December 18, 2024 at 11:18 PM
Giving local LLM a spin
Reposted by Ahmed Mansour
HOLY FORKING SHIRTBALLS!!
I just made my first SEO Katana sale! 😱
brb, crying...
I just made my first SEO Katana sale! 😱
brb, crying...
December 18, 2024 at 7:47 AM
HOLY FORKING SHIRTBALLS!!
I just made my first SEO Katana sale! 😱
brb, crying...
I just made my first SEO Katana sale! 😱
brb, crying...
Reposted by Ahmed Mansour
Live in 5 min 🔴
Today we’ll look at
- Marc Lou’s latest startup, Codefast
- ChatGPT’s $200/mo
- We find the best AI model for marketing
- Deskhub
- A guy that built 7 apps in 7 days
- Resend raises $18m series A
youtube.com/live/woF47Ft...
Today we’ll look at
- Marc Lou’s latest startup, Codefast
- ChatGPT’s $200/mo
- We find the best AI model for marketing
- Deskhub
- A guy that built 7 apps in 7 days
- Resend raises $18m series A
youtube.com/live/woF47Ft...
YouTube
Share your videos with friends, family, and the world
youtube.com
December 7, 2024 at 9:00 AM
Live in 5 min 🔴
Today we’ll look at
- Marc Lou’s latest startup, Codefast
- ChatGPT’s $200/mo
- We find the best AI model for marketing
- Deskhub
- A guy that built 7 apps in 7 days
- Resend raises $18m series A
youtube.com/live/woF47Ft...
Today we’ll look at
- Marc Lou’s latest startup, Codefast
- ChatGPT’s $200/mo
- We find the best AI model for marketing
- Deskhub
- A guy that built 7 apps in 7 days
- Resend raises $18m series A
youtube.com/live/woF47Ft...