๐ Python | ๐ค AI/ML | Software Engineer | Practical roadmaps โข Hands-on tutorials | Freelance & collabs open | DM ๐ฉ
Pune, India
- Tweets 66,924
- Following 666
- Followers 31,843
- Likes 33,842
Pinned Tweet
Building a finance AI agent that you can actually trust is a very different problem. You need more than an LLM and a clever prompt.
You need RAG, tool use, memory, reasoning, multi-agent architectures, evaluation, guardrails, observability, governance and a way to balance reliability, efficiency and cost.
That's what caught my attention about AI Agents for Finance.
'Building AI Agents For Finance' takes a hands-on Python approach to building financial agents with Claude and OpenAI models, and then goes much deeper.
You get into:
โข Agentic RAG and deep research
โข ReAct, reflection and other reasoning patterns
โข Multi-agent architectures
โข Fundamental analysis and trading systems
โข Insurance and compliance workflows
โข Agent evaluation and LLM judges
โข Tracing and observability
โข Guardrails and human oversight
โข Deployment, versioning and governance
What I like about this book is that it doesn't stop once the agent works. It gets into the questions that come next:
โข Can you evaluate it?
โข Can you observe it?
โข Can you make it reliable?
โข And what happens when things go wrong?
Especially in finance, those questions matter just as much as building the agent itself.
If you're working with AI agents, Python, finance or production AI systems, this is a book worth having on your radar.
You can grab a copy here link.amazon/B09iExZ7K
Paid partnership (ad)
27
8
78
5,904
What did you guys do this long weekend ??
An old photo I took during the Ganpati festival trip back home.
3
21
644
If you are an Indian student looking at pursuing MS Abroad, checkout
Paste your SOP and get a 0-100 score with line-level flags. Then paste the university list your consultant gave you and see an honest verdict on every school
truthpathms.com/
1
8
572
We turn text into tokens and embeddings for AI to understand it.
How does AI turn pixels into something it can understand?
4
1
14
854
There is a strange assumption in career development that you need to fix everything you're bad at.
So you make a list.
Communication.
Public speaking.
Leadership.
Technical depth.
Networking.
Time management.
Negotiation.
Then you spend years trying to turn every weakness into a strength. But not every weakness deserves that much attention.
Some weaknesses are worth improving because they limit your effectiveness. Others can simply be managed, delegated, or compensated for by building stronger complementary skills.
You don't need to become good at everything.
You need to become exceptionally useful at something while becoming competent enough in the areas that would otherwise hold you back.
3
1
28
1,056
How's your debugging flow post AI ?
Just hand LLM an issue and associated tools or actually try to work with an LLM/AI together?
14
20
1,195
Some of the most valuable things you do at work won't show up in a performance dashboard.
The incident you prevented. The documentation that helped someone solve a problem without you. The junior engineer you helped become independent. The messy process you quietly simplified.
There is no notification saying, "This made the team better."
And that's part of why experienced engineers sometimes underestimate this work. They're looking for visible accomplishments while their biggest contribution is making everyone around them more effective.
Not all career capital is measurable immediately.
Some of it compounds quietly.
4
1
14
992
This is a fascinating step toward making human-computer interaction feel truly natural.
AI needs to understand not just what we say, but how we communicate through expression, tone, and context, opens up some incredible possibilities.
Unlimited potential on this idea
Yesterday we previewed Griffin, a Human Interaction Model capable of seeing, hearing, sounding, and looking like a human does. There have been a lot of questions, so I wanted to take a moment to share our thoughts.
By way of introduction: Tavus is a research lab focused on enabling machines to meet us where we are, and to understand the nuances of how we communicate beyond words.
Griffin, our latest model, isnโt publicly available yet. Yesterdayโs announcement was a limited research preview to demonstrate what the model is capable of. With AI progressing so quickly, we prefer to share breakthroughs openly and in real-time as we work on a safe public release.
Face-to-face is how we evolutionarily communicate, it carries the most meaning and intent, and we want computers to be able to help with work that benefits from that emotional understanding, expression, and immersion. Some examples of the kinds of use cases we care deeply about:
- A tutor that can build understanding of how a student learns, see exactly when there is confusion or disengagement, and adapt the lesson to fit them.
- A health expert that can answer any questions about your upcoming appointment or prescription, at the pace you want, at any time you need, even on a weekend.
- A language coach that you can practice speaking with, that can correct your movement and pronunciation, and help build confidence to have real conversations
- Or the perfect assistant for everyone, that understands intent, knows how you work and remembers what matters.
Weโre working with partners on safeguards and systems for disclosure, as well as inviting discussions with officials around wider regulation and safe use. People will always know theyโre interacting with AI, while providing an interface that removes the need to โspeak computerโ.
We believe in a future where computers understand us well enough to make technology more accessible, more useful, and make us more capable as humans. That is the world we want to build, and we understand the responsibility to do so safely.
3
9
996
Jaydeep retweeted
Yesterday we previewed Griffin, a Human Interaction Model capable of seeing, hearing, sounding, and looking like a human does. There have been a lot of questions, so I wanted to take a moment to share our thoughts.
By way of introduction: Tavus is a research lab focused on enabling machines to meet us where we are, and to understand the nuances of how we communicate beyond words.
Griffin, our latest model, isnโt publicly available yet. Yesterdayโs announcement was a limited research preview to demonstrate what the model is capable of. With AI progressing so quickly, we prefer to share breakthroughs openly and in real-time as we work on a safe public release.
Face-to-face is how we evolutionarily communicate, it carries the most meaning and intent, and we want computers to be able to help with work that benefits from that emotional understanding, expression, and immersion. Some examples of the kinds of use cases we care deeply about:
- A tutor that can build understanding of how a student learns, see exactly when there is confusion or disengagement, and adapt the lesson to fit them.
- A health expert that can answer any questions about your upcoming appointment or prescription, at the pace you want, at any time you need, even on a weekend.
- A language coach that you can practice speaking with, that can correct your movement and pronunciation, and help build confidence to have real conversations
- Or the perfect assistant for everyone, that understands intent, knows how you work and remembers what matters.
Weโre working with partners on safeguards and systems for disclosure, as well as inviting discussions with officials around wider regulation and safe use. People will always know theyโre interacting with AI, while providing an interface that removes the need to โspeak computerโ.
We believe in a future where computers understand us well enough to make technology more accessible, more useful, and make us more capable as humans. That is the world we want to build, and we understand the responsibility to do so safely.
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
Itโs the first Human Interaction Model (HIM).
Community note
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard.
cellcog.ai/blog/tavus-griโฆ
research.nvidia.com/labs/amri/projโฆ
tech-ish.com/2026/10/02/tavโฆ
310
124
577
281,178
Do consider supporting my work here by subscribing
Its a small amount $1/month but motivates me to keep creating on this platform
x.lingyaoai.com/_jaydeepkarale/creatorโฆ
2
24
1,108
Jaydeep retweeted
โ๏ธ ๐๐น๐ผ๐๐ฑ ๐๐ผ๐บ๐ฝ๐๐๐ถ๐ป๐ด ๐ญ๐ฌ๐ญ: ๐ช๐ต๐ฎ๐ ๐๐ ๐๐, ๐ช๐ต๐ ๐ช๐ฒ ๐ก๐ฒ๐ฒ๐ฑ ๐๐ & ๐๐ผ๐ ๐๐ ๐ช๐ผ๐ฟ๐ธ๐ ๐งต
๐ ๐ช๐ต๐ฎ๐ ๐ถ๐ ๐ฐ๐น๐ผ๐๐ฑ ๐ฐ๐ผ๐บ๐ฝ๐๐๐ถ๐ป๐ด?
Cloud computing is the on-demand delivery of computing resources over the internet.
Instead of owning and managing physical infrastructure yourself, you rent the resources as and when needed from providers like AWS, Azure, or GCP.
โก๏ธ ๐ง๐ต๐ฒ๐๐ฒ ๐ฟ๐ฒ๐๐ผ๐๐ฟ๐ฐ๐ฒ๐ ๐ฐ๐ฎ๐ป ๐ถ๐ป๐ฐ๐น๐๐ฑ๐ฒ:
๐น Compute
๐น Storage
๐น Databases
๐น Networking
๐น Security
๐น Monitoring
๐น AI/ML services
๐น Application services
๐ ๐ช๐ต๐ ๐ฑ๐ผ ๐๐ฒ ๐ป๐ฒ๐ฒ๐ฑ ๐ฐ๐น๐ผ๐๐ฑ ๐ฐ๐ผ๐บ๐ฝ๐๐๐ถ๐ป๐ด?
โก๏ธ ๐ง๐ฟ๐ฎ๐ฑ๐ถ๐๐ถ๐ผ๐ป๐ฎ๐น ๐ถ๐ป๐ณ๐ฟ๐ฎ๐๐๐ฟ๐๐ฐ๐๐๐ฟ๐ฒ ๐ต๐ฎ๐ ๐๐ฒ๐๐ฒ๐ฟ๐ฎ๐น ๐ฐ๐ต๐ฎ๐น๐น๐ฒ๐ป๐ด๐ฒ๐:
โ High upfront hardware costs
โ Long provisioning times
โ Hardware maintenance
โ Difficult scalability
โ Data centre management
โ Capacity planning challenges
โก๏ธ ๐๐น๐ผ๐๐ฑ ๐ฐ๐ต๐ฎ๐ป๐ด๐ฒ๐ ๐๐ต๐ฒ ๐บ๐ผ๐ฑ๐ฒ๐น:
โ Need more servers? โ Provision them.
โ Need less capacity? โ Scale down.
โ Need infrastructure in another region? โ Deploy it.
๐ก ๐ง๐ต๐ฒ ๐ด๐ผ๐ฎ๐น ๐ถ๐ ๐๐ถ๐บ๐ฝ๐น๐ฒ: ๐๐ฒ๐ ๐ถ๐ป๐ณ๐ฟ๐ฎ๐๐๐ฟ๐๐ฐ๐๐๐ฟ๐ฒ ๐๐ต๐ฒ๐ป ๐๐ผ๐ ๐ป๐ฒ๐ฒ๐ฑ ๐ถ๐, ๐๐ฐ๐ฎ๐น๐ฒ ๐ถ๐ ๐๐ต๐ฒ๐ป ๐ฟ๐ฒ๐พ๐๐ถ๐ฟ๐ฒ๐ฑ, ๐ฎ๐ป๐ฑ ๐ฝ๐ฎ๐ ๐ฏ๐ฎ๐๐ฒ๐ฑ ๐ผ๐ป ๐๐๐ฎ๐ด๐ฒ.
๐ ๐ช๐ต๐ ๐๐ผ๐บ๐ฝ๐ฎ๐ป๐ถ๐ฒ๐ ๐ ๐ผ๐๐ฒ๐ฑ ๐๐ผ ๐๐ต๐ฒ ๐๐น๐ผ๐๐ฑ
โ Cost Efficiency
โ Need for faster time-to-market
โ Improve availability and disaster recovery
โ AI, machine learning, and big data workloads
โ Pay for use.
๐ก ๐๐ผ๐บ๐ฝ๐ฎ๐ป๐ถ๐ฒ๐ ๐ฑ๐ถ๐ฑ๐ปโ๐ ๐บ๐ผ๐๐ฒ ๐ณ๐ผ๐ฟ ๐ต๐๐ฝ๐ฒ. ๐ง๐ต๐ฒ๐ ๐ป๐ฒ๐ฒ๐ฑ๐ฒ๐ฑ ๐ณ๐ฎ๐๐๐ฒ๐ฟ ๐๐ต๐ถ๐ฝ๐ฝ๐ถ๐ป๐ด, ๐ด๐น๐ผ๐ฏ๐ฎ๐น ๐ฟ๐ฒ๐ฎ๐ฐ๐ต, ๐น๐ฒ๐๐ ๐ผ๐ฝ๐ ๐๐ผ๐ถ๐น, ๐ฎ๐ป๐ฑ ๐บ๐ฎ๐ป๐ฎ๐ด๐ฒ๐ฑ ๐๐ฒ๐ฟ๐๐ถ๐ฐ๐ฒ๐.
Whatโs been your biggest lesson moving to (or working in) the cloud? Drop it below. ๐
#CloudComputing #CloudEngineer #AWS #Azure #GCP #DevOps #SRE #Infrastructure #ksops #LearnInPublic
4
1
7
357
So someone has reported a copyright infringement against these two posts.
The fact that I have to defend this is already bad but the process to defend it is even atrocious
The reset password page on @X shows internal error even though the password is reset
When I submit the appeal the screen just gets stuck at "Preparing verification"
The level of respect wnd support this app has for creators is shocking @premium
11
22
1,402
Been stuck at this page since 25 minutes and this is after I submitted the appeal not before
So "preparing verification" makes no sense
2
2
203
Jaydeep retweeted
Building a finance AI agent that you can actually trust is a very different problem. You need more than an LLM and a clever prompt.
You need RAG, tool use, memory, reasoning, multi-agent architectures, evaluation, guardrails, observability, governance and a way to balance reliability, efficiency and cost.
That's what caught my attention about AI Agents for Finance.
'Building AI Agents For Finance' takes a hands-on Python approach to building financial agents with Claude and OpenAI models, and then goes much deeper.
You get into:
โข Agentic RAG and deep research
โข ReAct, reflection and other reasoning patterns
โข Multi-agent architectures
โข Fundamental analysis and trading systems
โข Insurance and compliance workflows
โข Agent evaluation and LLM judges
โข Tracing and observability
โข Guardrails and human oversight
โข Deployment, versioning and governance
What I like about this book is that it doesn't stop once the agent works. It gets into the questions that come next:
โข Can you evaluate it?
โข Can you observe it?
โข Can you make it reliable?
โข And what happens when things go wrong?
Especially in finance, those questions matter just as much as building the agent itself.
If you're working with AI agents, Python, finance or production AI systems, this is a book worth having on your radar.
You can grab a copy here link.amazon/B09iExZ7K
Paid partnership (ad)
27
8
78
5,904
