
22 | AI builder | i tweet a bit too much about AI x web3. | Maxi @CoinDCX
Lucknow, India
Joined September 2022
- Tweets3.3K
- Following838
- Followers776
- Likes9.4K
If I had 14 days to explore MCP for Web3
D1β2 β MCP fundamentals
D3β4 β Build a server
D5β7 β Add blockchain read tools
D8β10 β Connect an AI client
D11β14 β Test + deploy a demo
Connect models to useful tools.
4
If I had 21 days to build a smart contract audit assistant
W1 β Solidity + vulnerability patterns
W2 β Static analysis + LLM review
W3 β Test findings + build a report
AI assists audits. It doesn't guarantee security.
If I had 7 days to learn Foundry
D1 β Setup + forge
D2 β Write contracts
D3 β Unit tests
D4 β Fuzz testing
D5 β Invariant testing
D6 β Deploy to testnet
D7 β Publish the repo
Untested contracts are risky contracts.
22
If I had 7 days to build a blockchain chatbot
D1 β Choose a chain
D2 β Connect RPC
D3 β Fetch contract data
D4 β Add an LLM
D5 β Explain transactions
D6 β Test responses
D7 β Deploy
Useful beats flashy.
1
17
If I had 14 days to build an AI NFT explorer
D1β2 β NFT metadata
D3β5 β Embeddings
D6β8 β Vector search
D9β11 β Natural-language filters
D12β14 β Launch the explorer
Search beyond keywords.
22
Ritanshu Sharma retweeted
Up to $200K to move your startup to Abu Dhabi.
Pre-seed is enough to get in the door.
Hub71 (backed by Abu Dhabi's Department of Economic Development and Mubadala) brings tech startups to Abu Dhabi and helps them scale from there.
The package:
- AED 250,000 (~$68K) cash investment
- AED 250,000 (~$68K) in incentives for housing, office, visas and more
- Another AED 250,000 in cash for top performers after year one
What else you get:
- Housing
- Office space in ADGM
- Licensing and visa support
- Health insurance
- Incorporation costs covered
- Marketing, legal and financial support
- A 3-month Techstars mentorship track
- Intros to investors and corporate partners
Who it's for:
- Pre-seed, Seed or Series A tech startups from any country
- No registered company needed to apply
- Specialist tracks for AI, ClimateTech, Digital Assets and Life Sciences
The condition: one founder relocates to Abu Dhabi and builds a team there.
Apply link below π
hub71.com/program/access-proβ¦
Cohort 21 deadline: February 2027.
Bookmark this + tag a founder looking for a new market.
Ritanshu Sharma retweeted
If I had 6 months to master AI Cybersecurity Engineering.
I'd do this.
Stage 1: AI-Augmented Threat Hunting & Log Analysis
SIEM architecture, LLM log parsing, behavioral analytics, C2 beacon detection, anomaly scoring, automated triage.
Stage 2: Smart Fuzzing & Exploit Generation
AFL++, LLM-guided mutation, AST parsing, memory corruption, automated zero-day discovery, payload synthesis.
Stage 3: LLM Application Security (The New AppSec)
OWASP Top 10 for LLMs, prompt injection, multi-turn jailbreaking, Crescendo attacks, data exfiltration vectors.
Stage 4: Agentic Security & MCP Defense
Model Context Protocol security, JSON-RPC sanitization, confused deputy attacks, tool poisoning, blast radius containment.
Stage 5: AI-Driven Incident Response (Autonomous SOC)
SOAR automation, LLM reverse engineering, YARA rule generation, automated host isolation, incident timeline mapping.
Stage 6: Adversarial Machine Learning (AML)
Data poisoning, model inversion, evasion attacks (FGSM), adversarial training, gradient obfuscation.
Stage 7: Non-Human Identity & Agent Auth
OAuth2 for machines, SPIFFE/SPIRE, scoped JWTs, zero-trust for swarms, ephemeral credential brokering.
Stage 8: Securing "Vibe Coding" & AI Supply Chain
AI-generated code backdoors, hallucinated dependencies, AI-SBOM generation, Copilot/Cursor output auditing.
Stage 9: Privacy-Preserving AI & Cryptography
Differential privacy, homomorphic encryption, Trusted Execution Environments (TEEs), federated learning, weight extraction defense.
Stage 10: Automated AI Red Teaming
Microsoft PyRIT, Garak, LLM-as-an-Attacker, CI/CD security fuzzing, adversarial dataset generation.
Stage 11: Threat Intelligence & Dark Web Scraping
OSINT automation, Tor routing, NLP for threat feeds, credential leak correlation, dark web monitoring.
Stage 12: Public CVEs & Security Portfolios
Responsible disclosure, exploit chain teardowns, security advisories, open-source framework bug hunting.
Most people stay stuck watching tutorials.
Builders get hired.
(Bookmark it)
If I had 30 days to learn blockchain data engineering
W1 β RPCs + event logs
W2 β SQL + indexing
W3 β Subgraphs + APIs
W4 β Build an analytics dashboard
Raw data becomes useful insights.
1
19
Vibe coders, thank me later
21st.dev
designeer.xyz
reactbits.dev
magicui.design
uiverse.io
mobbin.com
inspora.design
2
23
Ritanshu Sharma retweeted
Remember the first time you made a trading mistake? π
What happened, and what did it teach you?
Share your trading story with us. The best story wins βΉ5,000! π
This Childrenβs Day, letβs learn from each otherβs experiences.
How to participate:
1. Follow @officialmudrex
2. Like this post.
3. Share your trading story in the comments.
π The most genuine story wins βΉ5,000!
Deadline: 15th November, 11:59pm IST
Turns out, some of the most important investing lessons are the same ones we were taught as kids.
Dont put all your eggs in one basket.
Be patient.
Dont let FOMO make your decisions.
We just gave these lessons a little makeover with The Greener Path, a moral science book by Mudrex for traders and investors.
And this Childrenβs Day, itβs finding its way into Bengaluru homes through @Swiggy Instamart orders.
Maybe weβre never really too old for moral science. π
#happychildrensday #mudrex #swiggy
If I had 14 days to build a DAO assistant
D1β2 β Governance data
D3β5 β Proposal indexing
D6β8 β RAG + citations
D9β11 β Voting summaries
D12β14 β Deploy a demo
Make governance easier to understand.
1
25
If I had 14 days to learn LLM evaluation
D1β2 β Build a test dataset
D3β4 β Accuracy metrics
D5β6 β Hallucination checks
D7β9 β RAG evaluation
D10β12 β Tracing + regression tests
D13β14 β Publish results
Measure before scaling.
1
20
If I had 30 days to explore ZK
W1 β Cryptography basics
W2 β Circuits + constraints
W3 β Proof generation
W4 β Build a demo with a ZK framework
Privacy is worth understanding.
2
30
vibecoders thanks me later π± ui.wensity.com
2
24
who made this ??!! π
peopleofbengaluru.com
1
22
If I had 21 days to build a DeFi AI assistant
W1 β DeFi + protocol data
W2 β RAG + risk explanations
W3 β Dashboard + testnet demo
Explain risks. Never promise returns.
1
19
If I had 7 days to learn LLM frameworks
D1 β Prompts + models
D2 β Chains + structured output
D3 β Tools
D4 β Retrieval
D5 β Memory patterns
D6 β Evals
D7 β Build an app
Understand the concepts first.
If I had 14 days to build a crypto research agent
D1β2 β LLM APIs
D3β4 β Market data APIs
D5β7 β RAG + source citations
D8β10 β Evals + guardrails
D11β14 β Live dashboard
Research needs evidence.
2
34
Pre-IPO trading used to be locked behind private equity funds and $1M net-worth accredited status.
Now, onchain perpetuals let you gain price exposure to companies like Anthropic before they hit public exchanges.
Here is how Pre-IPO perps work on @KriptoK. π #KriptoKPreIPO
1
3
85
Pre-IPO Perp Checklist Before Entering a Trade:
βοΈ Verified private valuation history
βοΈ Limit order priced near major support/resistance βοΈ Risk per trade capped at 1-2% of portfolio
βοΈ Stop-Loss and Take-Profit set concurrently
1
1
6
Trading private market valuations requires strict risk control and patience, but democratizes access to early-stage tech growth.
Which private tech company are you watching most closely for pre-IPO price discovery? Letβs discuss below.
Follow @KriptoK & @KriptoKLeague .
1
5









