chandrashekhar (@LazyLearner_) header
chandrashekhar (@LazyLearner_) profile picture

Co-Founder @ Deepvue

Bengaluru, India
Joined July 2010
Photo by chandrashekhar (@LazyLearner_)Photo by chandrashekhar (@LazyLearner_)Photo by chandrashekhar (@LazyLearner_)Photo by chandrashekhar (@LazyLearner_)
A compliance analyst can now ask Claude to onboard an Indian business — and Claude does it end-to-end. KYC, ownership, sanctions, court records, the DigiLocker consent moment with the director. Live today via Deepvue MCP ↓ deepvue.ai/mcp #MCP #Claude
1
1
2
4
620
chandrashekhar retweeted
Hiring a software engineer in Bangalore. We build identity, risk & fraud infra on India's DPI — APIs, an MCP server, a workflow builder. Decisioning is what's next. Python/FastAPI + agentic workflows in prod. 1–2 yrs. ₹12–15L. No cover letter: forms.gle/cRuimjEVyEqWqU6V7
1
1
25
What we'd be building: Identity, Risk, Fraud and Compliance for financial services, solved AI-first. We want someone as excited about those problems as we are. Apply -> forms.gle/2x8ucrSKCkfeR4J19
We're hiring Software Engineer at @godeepvue, in Bangalore. We build identity, risk, fraud, and compliance infrastructure on India's digital public infrastructure. Lenders, NBFCs, fintechs, and Internet Companies use our APIs and now our MCP server and workflow to verify customers and businesses at scale. That part is shipped and running in production. What we are building: identity, risk, fraud and compliance for financial services, solved AI-first. The work doesn't split into backend and AI. It's Python, FastAPI, Django and AWS, and it's GenAI and agentic workflows going from a notebook to something our customers depend on in production often in the same week. 1–2 years shipping production software. ₹12–15L. Hybrid, two days a week in the office. No cover letter. Tell us something you've built and shipped, with one line on what was hard about it. Apply → forms.gle/2x8ucrSKCkfeR4J19 What we do → deepvue.ai Founder → @LazyLearner_ #Hiring #Bangalore #Fintech #Python #GenAI #IndiaStack #NBFC
4
1
5
224
Three early usage patterns we're seeing: · Compliance — full lifecycle (onboarding, screening, DD) by asking · Ops & support — verification checks without pinging engineering · AI Builders — wiring Deepvue into agents(chat, voice etc.) and copilots
1
38
Status: · Live via Claude custom-connector today · OpenAI / Gemini support next · Pricing: per-check, wallet-based (same as our API)
1
45
Covered today: · KYC (DigiLocker/PAN) · KYB (GST/MCA) · PEP + sanctions + AML · Court records + FIR
1
29
The agent has to initiate a consent request, hand control to the human, wait, then pick the verified data back up — without losing context. Getting this right in an agent runtime is the harder half of the problem.
1
32
Consented flows are an under-explored shape in MCP. Most discussion so far has been read-only lookups. Anything that puts a user in the loop — payment auth, KYC, health records, Account Aggregator data — needs something different. We've been calling this Agentic Compliance.
1
36
Every existing MCP data connector — FactSet, S&P, Moody's — is server-side lookup. Data you already have rights to query. Indian KYC isn't that. Every pull needs the subject's real-time consent. That's a different architectural shape than request/response.
1
28
Background: Deepvue is compliance infrastructure used by forex, crypto, lending, and KYC teams. We've shipped KYC / KYB / AML APIs for years. The MCP server makes all of it agent-callable through Claude custom-connector. No new code — just connect Claude to Deepvue and ask.
1
60
As an early-stage startup, when you are double checking your outbound lead generation strategies and suddenly you receive two inbound leads. Everyone in the team
1
3
chandrashekhar retweeted
Newton never had to cold email someone. Else, he'd be more humble about the 3rd law.
2
1
35
With no insecurities at all. One needs sheer guts to do this. Thanks a lot! @sgaurav Surely I am learning from the best!😁
As a founder, it is a special feeling to see people moving on to start their own ventures. On one side, you are worried about losing one of your brightest but deep down, you root for them to make it big. @SatanAtWink @LazyLearner_ and Vishakha - win!
6
chandrashekhar retweeted
Been off for almost a week, have zero opinions on data or engineering anymore. It’s wonderful.
15
11
3
307
chandrashekhar retweeted
Welcome to our first episode of #TheTwentyMinuteMoat🙌🏼 Today, we talk to Sandeep Jindal - Head of #CustomerSupport @FURLENCO about "Furniture, WFH And How Customer Support Defines Culture." Click here to listen: https://teapawt.net/t.co/aNmumNWGhx🎙 #WFH #CustomerSupport
1
9
137
chandrashekhar retweeted
We're excited to announce our first PyTorch Developer Day starting 8 AM on November 12, 2020 PST. Learn more: pytorch.org/blog/pytorch-dev… #PTD2
3
118
23
551