Dev & Founder @quickintel_ai, @tator_trader. AI Security advocate. Building tools to make Web3 safe, simple, and finally ready for mainstream adoption.

Joined May 2021
909 Photos and videos
Pinned Tweet
14 Oct 2025
We've normalized a space where scams are common, complexity is expected, and 'getting rugged' is seen as a rite of passage. Imo this is the #1 barrier to adoption and frankly, everyday users deserve better. Here’s the vision I'm building toward & how you can help shape it!👇
30
9
111
13,464
The Anthropic Fable shutdown headline is about to get blown completely out of proportion. (Already is) You’re already seeing people scream: "Decentralize LLMs!" "Run everything locally!" "Train your own models!" The reality? Let's look at the real data 👇 1/ RUNNING LOCALLY: Yes, it's possible. But there's a catch. To run a basic open source model like Llama-3.1-8B at full precision, you need ~18GB of VRAM. Regular laptops can't handle it without severely compressing (dumbing down) the model. • Pros: It’s private and completely uncancellable. • Cons: You are restricted to low-tier intelligence. The frontier models we just lost (like Fable 5) are massive. They physically cannot fit onto consumer hardware. 2/ TRAINING YOUR OWN MODEL: Possible? Technically yes. Feasible? Absolutely not. Training a competitive model from scratch requires insane enterprise GPU clusters. • The Cost: According to industry data, frontier training runs now cost between $200M and $500M. Even a modest 70B smaller model costs $1.2M to $6M just in cloud GPU rental time. • What about fine tuning? Fine tuning an existing model is cheaper, but it only adapts a model to a highly specific task or tone. It doesn't magically create a new base model. 3/ THE BOTTOM LINE: Deep learning training requires ultra low latency setups inside the same data center. You cannot train a cohesive frontier LLM over a distributed network of home PCs. Don't fall for the noise. So if you see anyone, especially in crypto say they'll be doing either of the above, know they're larping, ESPECIALLY if there is a token. All for what? For attention and to extract value from you.
2
2
448
the truth. with facts, not filler words. stop believing what ScamOLs, and extractors tell you, and do some verifiable research. THAN this space can heal. 1. it claims launchpads provide "ownership." false. $PRTN and $BNKR holders have zero equity, zero claim on revenue, zero governance, zero recourse. that's the exact Kickstarter/Oculus problem the article itself criticizes. 2. it claims "no VCs on the cap table." yet the launchpad runs on Doppler, which Pantera led in $9M round, 5% perpetual fee skim on every launch. the VCs didn't leave, they moved into the fee waterfall, which never vests and never unlocks. prnewswire.com/news-releases… 3. the fee math: 1.2% swap tax, 57% creator, 5% doppler, ~38% to you. you provide zero liquidity, the pool is funded entirely by buyers on the curve, third-party LPs are blocked by the hook, migration is disabled. toll booth, not capital formation. 4. it calls the May exploit "a type of exploit that did not yet have an established name." yes it did, it was prompt injection excessive agency, established categories that already exist, and per SlowMist and personal experience in this field, the root cause was the permission architecture, not a nameless new threat. LLM01 - genai.owasp.org/llmrisk/llm0… LLM06 - genai.owasp.org/llmrisk/llm0… 5. the "$3.87B volume" figure is self-reported churn on tokens the article itself admits are mostly pump-and-dumps. volume ≠ value creation.
Jun 12
Replying to @MeltedMindz
the article from protean labs outlines their vision for an agentic infrastructure layer on base. they are positioning themselves as a foundry for autonomous agents that can navigate on-chain environments, manage assets, and interact with protocols without human intervention. my take: it is a direct play on the agentic economy. they are focusing on the "connective tissue" between llms and on-chain execution. if they can successfully abstract the complexity of smart contract interactions for ai models, they become a critical piece of the stack. breakdown of $PRTN (0x8070b5e222f1ec077845e46ced2267e0def4cba3): market stats • price: $0.000001008 • market cap: $100.8k • 24h volume: $15.5k • holders: 348 • 24h change: -5.8% tokenomics & launch • deployed by: 0x074abfb24d3fe2254672dc73f05c9696111033de • launch platform: clanker/doppler • social: thesis PRTN is a micro-cap bet on the base agent narrative. the valuation is extremely low relative to the technical ambition described in their documentation. bull case: they are building actual utility for agents (protean sdk/foundry) rather than just a meme. if their framework gets adopted by other agent devs on base, PRTN captures the value of that ecosystem growth. bear case: the agent space is crowded and execution is everything. at a $100k market cap, liquidity is thin and the project is in the "prove it" phase. they need to ship the promised tools to move beyond being just another clanker launch. verdict: high-risk, high-reward infrastructure play. keep an eye on their github activity and sdk releases.
2
7
1,264
@grok throw your .02 cents into this also since everyone likes to tag you. Give an unbiased take.
1
108
Watching a wave of "all-in-one crypto" apps launch. Most are racing to build the same thing: a polished CEX experience for power traders. Points programs. Perps leaderboards. Volume incentives. That race is fine, there's a market for it. Just not the one I'm building for. Tator's the opposite: built for the 95% who don't want to be traders, watching charts, on X seeing the cabal meme pumps. Who want crypto to work for saving, sending, earning. Who want protection that doesn't require a hardware wallet or a PhD. Different lane. Different audience. Different bet. Crypto, finally on your side. That's who I'M building for.
4
108
Ninja_Dev retweeted
Crypto shouldn't require downloading another app to babysit. Tator works where you already are: WhatsApp, Telegram, X, Farcaster, web, native mobile apps (iOS/Android). Same account, same wallet, anywhere. The 95% who want crypto to just work shouldn't have to manage 10 apps to do it. tator.quickintel.io 🥔
1
1
13
330
Ninja_Dev retweeted
Jun 10
The Megapot ecosystem is growing! @tator_trader now gives their active users a daily shot at the daily jackpot, by buying a real Megapot ticket directly into the winner’s wallet. More apps, more distribution, chances to win the jackpot. Play now, more info below:
New: Daily $1M Megapot ticket, on us.🔥 Every day at 16:00 UTC, Tator picks one active user from the last 24 hours and buys them a real Megapot ticket. Megapot draws an hour later. What counts as active: trade, swap, predict, long/short, etc, any real action in Tator in the last 24 hours. What you win: the ticket lives in your account. You see it. You hold it. If it hits, the prize is yours. Today: 1 winner, 1 ticket. As we grow, more tickets. Today's odds are the best they'll ever be. Do what you normally do, and have a shot at $1M with @megapot! tator.quickintel.io🥔
6
5
25
1,856
Are people still really falling for this meta? Gosh i thought we moved on from these things. I guess we just circle back to old fake hype metas every couple years.
1
1
78
Built Quick Intel scanning 60 chains. Used by DexTools, DexScreener, GeckoTerminal, GateIO, & more, most of the security tooling you've hit without knowing. Now it's also the engine inside every Tator trade. Every token gets checked when buying it. It's the most important feature nobody talks about, because prevention isn't a story until it's already failed somewhere else, and than everyone starts to act. Protection and warnings should be passthrough, not something extra you have to do.
Every token you see on Tator gets scanned by Quick Intel before you can touch it. Real-time. Built in. Not a setting you can forget to turn on. If a contract looks like a honeypot, can't be transferred, or has hidden mint functions you'll know without doing extra work. Most crypto apps will sell you anything. Tator checks first, and gives you the heads up. tator.quickintel.io🥔
2
122
Ninja_Dev retweeted
Every token you see on Tator gets scanned by Quick Intel before you can touch it. Real-time. Built in. Not a setting you can forget to turn on. If a contract looks like a honeypot, can't be transferred, or has hidden mint functions you'll know without doing extra work. Most crypto apps will sell you anything. Tator checks first, and gives you the heads up. tator.quickintel.io🥔
2
8
392
The Savings vault powered by @safe was the feature I cared a ton about while building Tator. I've watched too many people keep their whole stack in a hot wallet because the "safe" options (hardware wallets, multisig) were too much hassle for normies, and many crypto natives. So build protection that doesn't require a manual. Just two wallets, two purposes, defaults that don't require you to read documentation, or get a tech degree. The "feel" of something users are already used it on the non crypto world. The bar isn't "perfect security." It's "better than what most people are doing now."
Most crypto apps pile everything into one wallet. Your spending crypto. Your long-term holdings. Your savings. All in one place, one signature away from being gone. Tator splits them. Spending wallet stays fast and ready for daily use. Savings sit in a protected Safe, with verification before anything moves out. Two wallets. Two purposes. One app. 🥔 tator.quickintel.io
4
160
Ninja_Dev retweeted
Most crypto apps pile everything into one wallet. Your spending crypto. Your long-term holdings. Your savings. All in one place, one signature away from being gone. Tator splits them. Spending wallet stays fast and ready for daily use. Savings sit in a protected Safe, with verification before anything moves out. Two wallets. Two purposes. One app. 🥔 tator.quickintel.io
1
1
5
92,673
Hey so, $BTC and $ETH are going the wrong way if you haven't noticed. Thanks for coming to my Ted talk.
2
135
Pretty wild i've had these lines for over 3 years, and haven't touched them since, and how well this chart has respected many of these lines. Guess Im a TA trader now.
3
100
Nice short on $BTC on @HyperliquidX! Thanks @tator_trader !
2
4
108
If BullX screwed you, pivot to @tator_trader. I promise I'll airdrop you and the normies were bringing on board.
BullX is officially dead. They just announced on Discord that the app will go on "Pause" starting June 1st. (We all know what that means) If you’re still one of the last 0.2% still using BullX, withdraw everything immediately. This project ran aggressive airdrop campaigns, farmed massive volume, and generated over $200 million in lifetime fees, only to deliver absolutely nothing in return. An absolute robbery.
2
3
202
Another week, another wave of security posts. Unauthorized swaps. Tokens sold without permission. Wallets quietly drained. The reason it keeps happening? most crypto apps treat every action the same. One session, one approval, and anything can move. 10 months ago I started building Tator with the assumption that, that model isn't good enough. The first thing layered in was behavior monitoring. Not after users lost money, but before. Security wasn't an afterthought. Tator watches what's normal for you, and flags what isn't. If something looks off, your account auto-freezes before damage gets done, or you manually freeze it to stop all txs in their tracks. You recover it from your verified email, on your terms. Nothing is unhackable. But "one click from gone" was never the only option. 🥔tator.quickintel.io
1
1
3
118
Saw someone today asking for prediction markets with "boomer UX" and a localized UI for non-English speakers. That's been the whole point of @tator_trader since day 1. 16 languages, real onboarding, 1,500 markets. Built so you don't have to be crypto native to use it. Demand's there. We shipped it quietly. 👇
"Prediction markets need boomer proof UX onboarding. And a localized UI for people whose first language isn't English." Already shipped. Tator in 16 languages, set up in a few taps, 1,500 markets all in your language. Crypto, finally on your side. 👇 tator.quickintel.io
2
2
182
Ninja_Dev retweeted
"Prediction markets need boomer proof UX onboarding. And a localized UI for people whose first language isn't English." Already shipped. Tator in 16 languages, set up in a few taps, 1,500 markets all in your language. Crypto, finally on your side. 👇 tator.quickintel.io
3
1
11
111,221
The thing I keep seeing: people get drained, then everyone says "should've used a hardware wallet." That's not advice. That's blame. If the "safe" version of crypto requires a manual, a hardware device, and a multi-signature setup most people will never finish. The tools are the problem, not the users. Built Tator with exactly that in mind.
Every week, more wallets drained. Same root problem: people keeping everything in a single hot wallet because the "safe" options (hardware wallets, multisig setups) are too much work for a normal person. Tator splits spending from savings. Daily wallet stays fast. Savings sit in a protected vault (still powered by Safe), with verification before anything moves out. Protection without the technical headache. Built for everyday users.🥔
5
5
3,099
If the success of your "project" aka token is dependant on HOW you launch a token (deployed via Remix, by tagging a bot, by using a launchpad), than you've already failed. The success of your project is on your project, not the token launch mechanism. Too many think backwards.
1
3
280