Milind Pandey

Software Development || Web-3 || Finance || Capital Markets || Oration ||
In quest of wealth generating , problem solving juncture of Tech💻 n Finance💸

[Milind Pandey @themilindpandey]

Recently went through a lot of discussions regarding "onset of AI and the paradigm shift at work".

Conclusion drawn:
Pre AI fundamentals and Post AI speed is the moat one can/should have today.

[Milind Pandey @themilindpandey]

@Sheetal2205 i've been through AI assisted coding/debugging rounds (model under strict guardrails to not spit the code rather ask for approach, logic and give syntax) last month

as far as in person ones are concerned GCCs - jargons and how would one implement AI

[Milind Pandey @themilindpandey]

@AnandaniNisha i think platforms operating in IFSC at GIFT city provides tax optimised global access

[Milind Pandey @themilindpandey]

@safishamsii Headline: Stuck in traffic. Just like your SQL queries.

​Tagline: Relational databases weren’t built for connected data. We are.

[Milind Pandey @themilindpandey]

@Unnati_twts sorry, out of context but @failproofai and @neatlogs are in same space !? what differentiates you?
@simranrambles

[Milind Pandey @themilindpandey]

@CodeWithTamara start selling dev courses to desperate ones and you find yourself in a ready-pool of organic market for you :)

[Milind Pandey @themilindpandey]

@ananyairl i have polar take on this

1.initializing and staying with the conviction for long enough
2.abandoning sth which is not working out

[Milind Pandey @themilindpandey]

@ElitzaVasileva 1.2 mil coming in annually at 12% market appreciation would give me headroom to build, sell, travel and explore ppl, diversity, cultures thus attracting more money

so its a cycle u see!

[Milind Pandey @themilindpandey]

@waiting4_asi just out of curiosity, are your X replies automated ?

[Milind Pandey @themilindpandey]

i just rushed through building a ML pipeline for amazon ML challenge and 5.5 performed exceptionally well in reasoning around the blocking/candidate generation strategies taking my F_0.5 score from 0.61 to 0.73 at full run within 3 hrs (on a free tier colab - memory contraints)

[Milind Pandey @themilindpandey]

@waiting4_asi @abuxtalha i just rushed through building a ML pipeline for amazon ML challenge and 5.5 performed exceptionally well in reasoning around the blocking/candidate generation strategies taking my F_0.5 score from 0.61 to 0.73 at full run within 3 hrs (on a free tier colab - memory contraints)

[Milind Pandey @themilindpandey]

@waiting4_asi @thesachinray i disagree to the point that its the very reason for frustration especially in consumer grievance.

[Milind Pandey @themilindpandey]

@arpit_bhayani Would solicit a detailed take by you on Jev on YT/ preferably a stream where we leverage Jev and its classification capabilities hands on to build something.

I have a task (completed by agents) verification engine in my mind currently. @arpit_bhayani

[Milind Pandey @themilindpandey]

@moiiikaaa reasoning brotha reasoning , loyalty follows !

just came on X after 72 hrs locked in with a business entity resolution ML pipeline and Opus 5.5

[Milind Pandey @themilindpandey]

@safishamsii @graphify except for design and filmmaking all boxes ✅️

rather happy and convinced to see the roles existing where i thrive best with my multipotentiality !
Techie (not nerd) + people's person (oration , my forte)

would love to add value!

[Milind Pandey @themilindpandey]

<Amazon ML Challenge>

done with iteration 3 of the complete ML pipeline boosting F_0.5 score :
0.61 -> 0.65 -> 0.72

vectorisation played a key role from 2 to 3.
wrapping up 72 hrs of grind with tons of knowledge and skill-ups pertaining to ML pipelines.
@amazonIN
@awscloud

[Milind Pandey @themilindpandey]

<Amazon ML Challenge>

finaly after 54 hrs of sleuthing ....
done with the submission of matching.tsv output

12.7 GB RAM, vectorizations, blocking optimisations, etc.

onto some feature engineering, threshold optimizations for remaining time
@awscloud @amazonIN
#MachineLearning

[Milind Pandey @themilindpandey]

@mayuri_3015 60 dollars can get you 2 claude pro instances and 2 opencode go instances - never ever you will run out of time window

just manage context through md files.

[Milind Pandey @themilindpandey]

compute is the bottleneck!

exhausting token limits mid-project feels as if we are pre 2016 and our 1 GB data pack got exhausted :)

Working out Blocking strategies that work at scale - SNB failed.

#AmazonMLChallenge
#MLTraining
@awscloud @amazon @amazonIN

[Milind Pandey @themilindpandey]

@kirat_tw ig finance is the biggest space/basket these regulations and products like @ZeroDriftSec pose their importance