Challenge for FSD: Load a few 5-gallon water jugs in the Model Y trunk. Drive without them falling over.
Humans do this easily. Every current mode accelerates and turns too hard.
Add a new Chauffeur Mode that passes.
@elonmusk@Tesla
@Erdayastronaut - Where was the notice about the NASA / Intuitive Machines press conference?
https://t.co/fqbrG5lQjv
https://t.co/QiTdbKIW5s
Starts in 4 minutes - 5PM Eastern
@karpathy theory time!
What if he is leaving to start the AI/ML department for @elonmusk TITS University?
1. Andre gets to teach and create the curriculum
2. Elon gets a steady flow of engineers who are productive on day one
3. Students have guaranteed employment
@karpathy Can we get back to self-hosting and the way the "inter"net was architected?
It would be awesome if you could have peer based backup/mirroring. Essentially host 2-3 friends data and they host yours (encrypted backup of files, email, etc.) Have a falling out? Revoked
@Apple I'm specifically talking about the windows on monitor 1 moving to monitor 3 or any other screen. I have 4 external monitors on Sonoma 14.3. 3 Dell S3221QS and a single LG UltraFine.
@Apple Macbook Pro window positions.
Anyone else use their laptop in clamshell with external monitors (more than 1) and have the windows change screens every time the laptop sleeps and you re-login? It is so annoying to have to move the window positions EVERY single time.
🚀 Rethinking Tech Interviews: Beyond Data Structures & Algorithms 🚀
FAANG companies have become standard-bearers for tech interview practices. But do data structures and algorithm challenges truly uncover the top team players, or do they risk overlooking hidden talent?
With the rise of companies like @LeetCode, preparing for tech interviews feels like studying for the SATs. But once on the job, it’s quick adaptation, effective communication, and continuous improvement that often count more than algorithmic wizardry.
Having worked with startups, I’ve seen speed and adaptability take precedence. Scaling isn’t just about writing code - it’s about optimizing processes for peak efficiency, whether that's through UI improvements or strategic tech stack adjustments.
A personal confession - I flunked my technical interview with @BirdRide, but they hired me, perhaps seeing potential beyond that one interview. As a potential technical interview with @anduriltech looms on my horizon, I’m reflecting on what interviews really tell us.
As a founder, CTO or a hiring manager what are your insights?
* What has been your golden metric?
* What makes an effective tech interview?
* How can we do better in matching skills with roles?
@STLChrisH Curious how you track these subscriptions, invoice customers, track sms and phone calls. Custom software, CRM plus add-ons, manual (spreadsheet) or something I’m not thinking of
@isaiah_p_taylor I've been thinking about how to pivot some productivity tools I built and this might be it!
SMS is great to ingest thoughts (and most phones have speech to text) or MMS for raw audio and use Whisper on the back end.
Showing/finding notes (RAG) will be most difficult part.
Is there a market for a privately hosted productivity suite? Email, calendar, tasks, document storage, analysis, ML+LLM analysis etc?
Being a developer who thinks privacy is extremely important, I would think so... But, how many people are willing to jump through hoops to install/host it? Or even have the hardware at home to run the models?
License would be free for personal use, commercial fee based? Follow popular opensource software and offer hosted solution?
Bring your own LLM or use @OpenAI@AnthropicAI etc?
Host data on AWS, MINIO, DigitalOcean, GCP, iCloud?
Speech to text and text to speech (self hosted and/or cloud products)?
Thinking out loud...
@BrianRoemmele Unfortunately NeuSpeech reported a problem with their evaluation on 2023-11-29 and the misleading results.
"We have written a corrected version to use model.generate to evaluate the model, the result is not so good"
https://t.co/dRCqfMFNKy
Productivity tools
What do people really need to improve their workday without causing MORE work?
I built a platform to manage contacts/address book, emails, txt messages, phone calls, tasks and calendars (currently adding invoices with #stripe payments) This was created to allow virtual assistants to help small businesses in the construction industry (HVAC, Plumbing, Electrical, GC's, etc.)
These businesses need the help and don't want to learn new technology, the goal is to help generate invoices, accounts receivable, keep clients informed and assist with rescheduling. All this while we are just a phone call, email or txt message away.
It has been very difficult to change their behavior and delegate tasks. Pivot or push forward?
Features:
* Real time transcription and recording for phone calls with AI generated summary on conclusion
* AI writing assistant for email and txt messages (#GPT)
* Amazon #Textract to extract text from images and PDF's sent via MMS or email.
* AI based tagging for all documents sent to the assistant via MMS or email (ex. Invoice, Receipt, Packing List, Bill, etc.)
* Document association to contact(s) in the address book
* AI based document summation
Planned additions:
* Invoicing with stripe payments (Credit Card and PCH via connect product)
* #RAG (Retrieval Augmented Generation via to allow "chatting with your business")
* Real-time speech to text -> LLM -> text to speech with low latency on top of RAG and custom functions. This should allow creating invoices, sending SMS or emails, forwarding documents, scheduling/rescheduling appointments and finding documents such as warranty and payment status.
Questions?
* Should the platform pivot to direct to customer? (currently targets virtual assistants)
* Focus on a chat based UI? (on top of the existing features)
* Something completely different I'm not thinking of?
#Productivity #ML #OpenAI #LLM #Tasks #Calendar #VirtualAssistants #payments #Transcripts #Documents #constructionworker #assistant #Building