Dhruv Sharma

MAC Research MSCI. Macroecon, Stat. Phys. and Everything in between. Previously @FabricRisk @Polytechnique, @ENS_ULM. Views own. RT != Endorsements

[Dhruv Sharma @DhruvComplex]

Just as the Nobel Committee recognizes the importance of complex systems research today, we at @FabricRisk are building upon these insights to build a platform that captures the dynamics of financial risk in all its richness and complexity.

[Dhruv Sharma @DhruvComplex]

Today is a great day for the Complex Systems community. Giorgio Parisi has been awarded the Nobel Prize "for the discovery of the interplay of disorder and fluctuations in physical systems from atomic to planetary scales. "

Lets unpack that a little 1/n

https://t.co/UUviwq05NF

[Dhruv Sharma @DhruvComplex]

@drmtgr This can’t be real.

[Dhruv Sharma @DhruvComplex]

@drmtgr @gine_robert Tu le cherches aussi hein? Et puis c’est dans l’air des temps. Les boomers piquent déjà l’argent, pourquoi pas des petits fours ;)

[Dhruv Sharma @DhruvComplex]

This hits hard. I see a lot of myself in this as well. Lots of ideas that seemed interesting that I have worked on by myself. But always second guess and not share with the world.

Lesson: just do the work. Release it. And forget about it.

[Dmitrii Kovanikov @ChShersh]

I'm going to be vulnerable for a moment and admit that working on your own project is hard.

Unless you have an extreme belief in your vision, fighting self-confidence issues takes a Sisyphus toll on you.

It's not imposter syndrome, "Am I good enough?"
It's more like, "Does the world really need this?"

Only if you're an oracle or have a ton of experience can you predict what will be a banger. Most of the time, it's just dumb luck. Or right timing. Or leveraging hype waves. Or recycling past success when everyone has already forgotten about it.

But great execution could also be devastated by misfortune equally well.

When you work with someone else, and everyone believes in a project, at least you have some indication that more than one person might actually like it.

When you're alone, you might as well be the only person who cares about what you do.

[Dhruv Sharma @DhruvComplex]

@goyal__pramod Where is this from ?

[Dhruv Sharma @DhruvComplex]

This is the way!!

[JB Rubinovitz @rubinovitz]

Build a graph schema in Postgres and switch to a graph database when you really need to.

(You never will)

[Dhruv Sharma @DhruvComplex]

@drmtgr Ooh!! You guys are too good. My thing would be orthogonal and very much limited in scope. (For reasons you know well )

[Dhruv Sharma @DhruvComplex]

Absolutely! Working on supply chain risk at the moment. Knowledge graph construction is essentially a network building operation. Much better to use the network properties for traversal rather than use something like a dedicated graph database.

[jason @jxnlco]

why i avoid graph databases

the question: "is knowledge graph rag production ready? should we use it?"

the answer: after 10 years in ml, i stay away from graph databases. every company i've seen go into the graph world moves back to sql within 4-5 years.

the issues are real:

hard to hire talent (easier to find postgresql experts)
schema definition creates endless debates without clear best practices
most use cases need only 1-2 traversals, not complex graph operations
even facebook's "graph" was actually a large mysql database. the only company that truly needs graph databases is linkedin for 3-5 degree friendship calculations.

even for microsoft's document graph approach - i'd rather use fine-tuned embeddings. a graph is just an adjacency matrix, and fine-tuning can get you close to that similarity definition without the operational complexity.

start with your data: let specific use cases justify graph complexity rather than choosing technology first. graph might be 2% better, but traditional approaches working well means that 2% rarely justifies the maintenance cost.

[Dhruv Sharma @DhruvComplex]

@drmtgr This is another thing I am working on right now. Hoping to put an arxiv paper out (subject to internal … review)

[Dhruv Sharma @DhruvComplex]

@drmtgr What do you mean? Doing supply chain ?

[jason @jxnlco]

why i avoid graph databases

the question: "is knowledge graph rag production ready? should we use it?"

the answer: after 10 years in ml, i stay away from graph databases. every company i've seen go into the graph world moves back to sql within 4-5 years.

the issues are real:

hard to hire talent (easier to find postgresql experts)
schema definition creates endless debates without clear best practices
most use cases need only 1-2 traversals, not complex graph operations
even facebook's "graph" was actually a large mysql database. the only company that truly needs graph databases is linkedin for 3-5 degree friendship calculations.

even for microsoft's document graph approach - i'd rather use fine-tuned embeddings. a graph is just an adjacency matrix, and fine-tuning can get you close to that similarity definition without the operational complexity.

start with your data: let specific use cases justify graph complexity rather than choosing technology first. graph might be 2% better, but traditional approaches working well means that 2% rarely justifies the maintenance cost.

[Dhruv Sharma @DhruvComplex]

@MajinBoson Plotting a variogram would do it non? I think the formula for that would be in one of JPs books.

[Dhruv Sharma @DhruvComplex]

@MajinBoson @TokenOfTheMonth Yeah some of the numbers, even from YC funded startups, are not making much sense. Low base salary is expected for startups but founding engineer (like 1st hire, 2nd hire) and the equity is also super low.

[Dhruv Sharma @DhruvComplex]

@drmtgr @thesard1319 Je pense que “how china escaped shock therapy” de Isabella Weber montre comment la chine a pu devenir le super pouvoir. C’était pas totalement planifié ni totalement capitaliste.

[Dhruv Sharma @DhruvComplex]

@MajinBoson @alz_zyd_ I really don’t see the point of them. They seem to have become a very important part of recruiting in some circles it seems.

[Dhruv Sharma @DhruvComplex]

@ayushchopra96 @_R4V3N5_ We should connect @ayushchopra96 Working on a large scale simulation for finance.

[Dhruv Sharma @DhruvComplex]

Very interesting. Will software engineers in the age of AI finally merit the title of “engineers”. ?

[Drew Breunig @dbreunig]

Was recently talking with @jxnlco and @jeffreyhuber about how AI requires engineers to deal with probabilistic functions, and how data scientists are perhaps better prepared for building with AI…

On that note, I’ll be thinking about this HN comment for awhile…

[Dhruv Sharma @DhruvComplex]

@btaylor Couldn't agree more. I call it persona based testing: https://t.co/k7rouQrEpQ

[Dhruv Sharma @DhruvComplex]

@remilouf Mirrors my experience as well. I tried “yolo” mode with Claude code with tight guidelines on using TDD(red /green refactor). Still failed.

[Dhruv Sharma @DhruvComplex]

@drmtgr @thechosenberg J’imagine. Y a un moteur de croissance ici qui n’existe pas (plus) en France.