IT,Tech,LLM,AI,NFL,music,fitness,positivity.
Open to professional & common topic connections; intentional about long-term, meaningful personal connections.
Career learning: please do this weekly:
Meet Boss 1:1
Wins / impact
Risks and political/organizational observations
Relationships
30-60-90 progress
Learning
Weekly job application + JD + learning plan + 50+ slide deck
Lessons learned
Priorities for the coming week
@ConsciousRide@devXritesh Why not just check what it is supposed to do in the expected amount of time and tokens? Seems way easier
If the expected state of the target object has not been reached within 2 min or 1000 tokens,abort and send alert?
@aravind Jensen is far too important and knowledgeable to be sacrificed for a private affair. Most knowledge of IT nowadays is not a secret, chip building included
@suraj_sharma14 Oh yeah, I take your points and add to my learning goals.
These at enterprise scale are each a few week worth learning. AI engineering interviews are almost a deep dive on the topic you claim to have worked on.
I interview and have been interviewed. 🙂
If you are a Data Engineer, here are 4 deep-dive articles worth saving.
Real experiments. Real numbers. Spark UI observations. And lessons you can use when things go wrong in production.
Save this post. You might want to come back to these.
- What happens when you read 40 Billion Records with Databricks Disk Cache?
https://t.co/cfejIEdn69
- How did AQE reduce a Spark job from 45 minutes to 21 minutes?
https://t.co/iCXXfEQ2Dz
- Parquet is already fast. So why does Delta Lake exist?
https://t.co/2L6ZLgQqHY
- Why did 29,200 small files become a performance problem for Spark?
https://t.co/nRbWn7YLaN
And the next experiment?
A Spark join that refused to get faster even after adding more compute.
That's what I'll break down next.
Until then, what production Spark problem should I experiment with next?
@DasCapital8 I had not even seen a computer until my first day in master. 😂
I turned out OK.
A solid Maths and Physics guy throughout!
You are absolutely right.
@prakdadlani 1 and 2. Not just for business
Money almost never goes to ppl who chase it.
Almost all self made rich people
obsessed with following their passion to create a product that can be used.
Money just followed them.
Bill Gates, Larry Ellison, Steve Jobs.
CANCEL your weekend plans.
The 2027 meta just dropped. You NEED to:
• Build a computer-use agent that parses the DOM, diffs screenshots and maps clicks to coordinates so it can drive legacy software with zero APIs
• Ship generative UI where your LLM returns live React components instead of markdown blobs and watch your product feel 10 years newer
• Wire a System-One router that resolves 95% of requests in milliseconds for fractions of a cent, escalating only the hard 5% to reasoning models
• Kill your `while True` loop and rebuild it as a Temporal state machine that resumes from the exact node after a crash or a 3-day approval
• Write a raw MCP server in pure JSON-RPC so any external agent can call your product with scoped credentials, no SDK required
• Add grammar-constrained decoding so your agent emits 100% valid JSON on the first try, every try, forever
• Build a trajectory grader in CI that blocks the merge the second a tool-call sequence deviates from the golden path
• Ship an inter-agent sanitization proxy that executes prompt injections before they cross a single agent boundary
• Wire a PII unmasking proxy that swaps sensitive data for UUIDs before inference and restores it after with full audit logs
• Add a cost kill-switch that halts any loop projected to exceed $0.50 and degrades frontier → mid-tier → local → cache
• Implement KV-cache prefix sharing at the gateway and cut Time-To-First-Token by 80% on repeated system prompts
• Set up speculative decoding: local 3B drafts, frontier 70B verifies, acceptance rate logged on every single request
• Shadow-route 5% of production traffic to your new model, diff trajectories and cost silently, auto-post the report to Slack
• Automate the DPO flywheel: thumbs-down → preference pairs → nightly LoRA → eval-verified promotion, zero human touch
• Build a voice loop with VAD and sub-1s interruption handling, streaming tokens and tool calls simultaneously
• Ship an API-to-browser fallback that spawns a headless browser the moment a REST endpoint 404s, finishing the job anyway
• Add a WebGPU shim so a 2B model runs in the browser when the cloud dies and your users never see a spinner
• Fuzz your own agent swarm with malformed tool outputs and context overflow until it self-heals without a single pager alert
• Publish one benchmark: your system vs the naive baseline, on cost, latency and task success, fully reproducible
The wrapper era is over. The systems era pays.
You have way too much to do.
Bookmark & Repost.
@PerfInjust@Avik@CharlesFLehman There is nothing called mass migration.
Number of people migrating to America has been frozen at one million since 1990
It’s shameful and hilarious that you do not know how immigration works and have 35K followers.
Ethnic? Even more idiotic.
7% limit per country.