1. Collaborate strategically with your AI; don't just prompt.
2. Always plan WITH your AI before it codes.
3. Proactively manage the AI's context window.
4. Use capable models for complex, agentic work.
5. Give your AI persistent knowledge through Rules Files & Memory Banks.
Excited to release the 1st Indus Valley Annual Report to the world.
Indus Valley = our moniker for India's startup ecosystem.
The 1st issue gives us a chance to take stock of its rise and evolution, take a look back at 2021 & finally look ahead to what is coming.
@amal_vats
Our new AI system learned speech recognition in English with *zero* speech to text training data: researchers just gave it lots of audio, and it figured out what the words were. But it goes way beyond that - it learned Swahili too!
Nice writeup by @Analyticsindiam on why I think we should move from model-centric AI development (where the emphasis is on improving models) toward #DataCentricAI (where we systematically improve the data, using MLOps tools). https://t.co/7ClnrLKVGC
Bruce joins President @barackobama for a long and meaningful conversation that touched on so much of what we’re all dealing with these days. Listen to the first two episodes of their new podcast Renegades: Born In The USA now on @spotify. https://t.co/D7hvKyCxxQ
Engineers who went into mngmt - and back - really are amazing.
Some of the best staff engineers I know did exactly this: they can now balance so much easier between business & technology. Better mentors & “translators”.
As an industry, we should encourage this path far more.
@johncutlefish Controlling and evolving entropy in terms of tech ( languages, frameworks, tooling). Limiting the number of moving pieces at different scales
Few eng orgs have clear engineering strategies, they can feel almost mythical & intimidating to write. While struggling myself, I’ve landed on a reproducible way to write useful strategies: five 5 design docs, synthesize them into 1 strategy, repeat!
https://t.co/ewAaxq9aym
I'm looking for some references on continuous, safe deployment of statistical/ML models in production. Particularly interested in the model evaluation piece -- how do you know when a new model is good enough to replace an old one, and can you trust an automated decision here?
A decade+ ago, when music streaming went online, it seemed like not having ads or radio hosts were a feature. We had full control over the songs w/o interruptions. But I kind of miss talk show radio, listening didn't feel as lonely. So what happens if we merge music and podcasts?
I’ve reviewed over 750 PR’s at AWS. As my team’s tech lead I provide insightful feedback and enforce a high code quality bar.
But as a jr engineer I couldn’t review code. I didn’t know where to start, what to look for or how to comment.
Here’s how I review PR’s. 🧵 👇
When I first started my career, I thought being a stellar software engineer means writing correct, performant, and elegant code, having deep understanding of programming languages & frameworks & operating systems & networking & databases, etc.
@mfloryan Last name isn't a thing where I grew up 🙂. Trouble of not having last name on my passport was way too much so I ended up changing part of my first name as last name on my pass 😆