MojoAnalytics's profile picture. POV at intersection of #Fintech #Banking #BaaS #AI #Reg #Digital. Analytics guy with strong opinion. 

Listen more. Talk less. Full time geek.

Aditya Khandekar

@MojoAnalytics

POV at intersection of #Fintech #Banking #BaaS #AI #Reg #Digital. Analytics guy with strong opinion. Listen more. Talk less. Full time geek.

Aditya Khandekar reposted

How can #GenAI be safely deployed in large financial institutions and mitigate unintended risks? Introducing Project GGX, our joint Gen AI initiative with Corridor Platforms, that aims to address these key questions > owy.mn/3Vrxgna

OliverWyman's tweet image. How can #GenAI be safely deployed in large financial institutions and mitigate unintended risks? 

Introducing Project GGX, our joint Gen AI initiative with Corridor Platforms, that aims to address these key questions > owy.mn/3Vrxgna

Just visited the crucible of Hindu religion at ayodhaya, sharing this beautiful reel. Gratitude and best wishes for 2024 to everyone.


Great trip to Dubai... The theme of family centric living is very ingrained in middle east, gives hope in these turbulent times.


Just 26% of Gen Z and 14% of millennials use CU, as per GOBankingRates survey. CU's need to upgrade digital decisioning tech, analytical and related processes to stay competitive! at.law.com/KRZHg1?cmp=sha… via @CookeonCUs


India landed the lunar probe Vikram from Chandrayaan-3 on the south pole! What a fantastic technical achievement and so PROUD!

MojoAnalytics's tweet image. India landed the lunar probe Vikram from Chandrayaan-3 on the south pole! 

What a fantastic technical achievement and so PROUD!

Based on the Fed's Q2'23 report, frequency of delinquent payments are higher (chart), but, consumers are experiencing less financial distress. The unknow lever is student loan debt, which at 9.2% of all consumer credit has a deadline as payments are set to resume in Oct'23.

MojoAnalytics's tweet image. Based on the Fed's Q2'23 report, frequency of delinquent payments are higher (chart), but, consumers are experiencing less financial distress. 

The unknow lever is student loan debt, which at 9.2% of all consumer credit has a deadline as payments are set to resume in Oct'23.

Aditya Khandekar reposted

Weekend cook up ........ when #recsys meets #Llama2 ... PS, @huggingface 's TRL library is soo cool ... it just boils down to the dataset format to get started with baseline fine-tuning ... blog coming soon 😁

anindyadeeps's tweet image. Weekend cook up ........
when #recsys meets #Llama2 ... 

PS, @huggingface 's TRL library is soo cool ... it just boils down to the dataset format to get started with baseline fine-tuning ... 

blog coming soon 😁

Very proud of the work we have done here @AnindyadeepS Bottom line: Building enterprise use cases with LLM's require balancing speed and benefit against risk of going AWOL! There are best practices to manage risk... we will present more in the next blog in the series.

Domain specific LLM fine tuning always confused me. After several iterations I along with @MojoAnalytics published our first blog on best practices and insights on fine tuning LLMs using @huggingface and managing experiments using @weights_biases 👉 #LLM link.medium.com/m6yt5KPPZBb



Quick GAI insight: When building highly constrained categorization use cases (ex: categorizing text to issue), consider using Encoder Transformer with supervised learning on top. We got very robust & stable results! Right tools for right tasks😀 @LangChainAI @TaoWang2218


Nice chart showing economic expansion duration since 1945, the LONG outlier being the 2009-2020 period of 11 years. Soft Landing? Not sure yet, 2nd half of 2023 will tell.

MojoAnalytics's tweet image. Nice chart showing economic expansion duration since 1945, the LONG outlier being the 2009-2020 period of 11 years. 

Soft Landing? Not sure yet, 2nd half of 2023 will tell.

Aditya Khandekar reposted

Wondering what these jargons like #knowledge base, and #vector DB's knowledge retrieval are? How these databases like @pinecone , @trychroma work? How it helps #LLMs to tear down #hallucinations? Check out my latest blog 👉 link.medium.com/JOZ2vfp2QBb


Most of 2Q'23 earnings are in. Deposits held steady, no bank runs thankfully post results. Banks are now focusing on improve NIM's (profits) by shedding low yielding assets in their portfolio. Also helps shoring up capital. Any thoughts from banker friends here?

MojoAnalytics's tweet image. Most of 2Q'23 earnings are in. Deposits held steady, no bank runs thankfully post results.

Banks are now focusing on improve NIM's (profits) by shedding low yielding assets in their portfolio. Also helps shoring up capital.  

Any thoughts from banker friends here?

Consumer sentiment back at Q321 level, VIX low, put-call ratio low, equity markets on a tear... .... we were in similar position in late 2021, market peaked and then dropped 19% in 2022.

MojoAnalytics's tweet image. Consumer sentiment back at Q321 level, VIX low, put-call ratio low, equity markets on a tear...

.... we were in similar position in late 2021, market peaked and then dropped 19% in 2022.

Updated up to Q1'23, this FDIC chart for Net Interest Margin shows the compression trend. Blended cost of funds around 1.6% with rising rates & expectation from depositors and tapering yield on securities & loans are causing it.... .... 2Q'23 bank earning will reveal more.

MojoAnalytics's tweet image. Updated up to Q1'23, this FDIC chart for Net Interest Margin shows the compression trend. 

Blended cost of funds around 1.6% with rising rates & expectation from depositors and tapering yield on securities & loans are causing it....

.... 2Q'23 bank earning will reveal more.

A key insight from deep analytical work on GAI from us. Not all foundational LLM models are equal when using them for use cases. Foundational LLM will specialize as more vendors enter the fray catering to specific family of use cases, also by verticals. @AlexH_Johnson


While we know this adoption of GPT as 'cliche'... note... what got the 1st wave adoption will not drive the 2nd wave adoption, where use case will be more nuanced, embedded in broader workflow & need to de-risk from privacy, hallucinations, regulatory etc. concerns.

MojoAnalytics's tweet image. While we know this adoption of GPT as 'cliche'...    

note... what got the 1st wave adoption will not drive the 2nd wave adoption, where use case will be more nuanced, embedded in broader workflow & need to de-risk from privacy, hallucinations, regulatory etc. concerns.

Sourced from Bain Capital, nice summary of contrast between Traditional AI and GAI. Banking is in the 'toe-dipping stage' of adoption. @Clagett @rshevlin @commbankerguy

MojoAnalytics's tweet image. Sourced from Bain Capital, nice summary of contrast between Traditional AI and GAI. 

Banking is in the 'toe-dipping stage' of adoption. 

@Clagett @rshevlin @commbankerguy

Sometime an image is worth thousand concepts... this one is batting above avg. ! Recent market rally is broad based beyond the 'fantastic 7' as reported in WSJ. Are we in for a soft landing, thoughts from my network?

MojoAnalytics's tweet image. Sometime an image is worth thousand concepts... this one is batting above avg. !  

Recent market rally is broad based beyond the 'fantastic 7' as reported in WSJ. 

Are we in for a soft landing, thoughts from my network?

Are we atrophying human neurons for logical reasoning and problem solving by using the crutch of ChatGPT? I was tempted to summarize a HBS case study today with ChatGPT versus deeply understand it myself and addressing questions Who else sees this as a problem?


I would love to get a view of the Community FI's bankers here on this WSJ article on the impact of rising interest rates. @commbankerguy @JillCastilla @ICBA wsj.com/articles/every… via @WSJ


Loading...

Something went wrong.


Something went wrong.