←
kaledora5,596 posts
kaledora avatar✓

kaledora ✓

@kaledora

building @ostium | prev harvard, ballerina @ royal danish ballet

◷ Joined October 2020▧ 543 Photos and videos
kaledora avatar
kaledora@kaledora·21 May 2024

19/ TLDR; perps will become the listing engine for RWAs, outpace tokenizers, and usher in the golden age of the 'perpification of everything' disagree with me in any comments below :)

Abdul R avatar
⟳ kaledora retweeted
Abdul R@AbdulFromQueens·Nov 13

Over the last few months, we’ve been building infrastructure for the age of AI at Ostium. It’s not an easy decision for a startup that is finding success in their product to decide to slow down and rebuild their infrastructure from scratch. But that's the call that @kaledora and @contrarianmarco made. The hack on the legacy infrastructure was gutwrenching. We focused our efforts and energy on the migration, resuming trading, and continuing to build a world-class foundation. The technical team at Ostium has been quietly investing in some of the most ambitious tech I’ve seen in a long time. We can’t wait to unveil V3, and until then, we’ll have two more articles for you talking about our core systems.

kaledora avatar
kaledora@kaledora·Nov 13

Today we're publishing the first article in our security series. It's the first time we've explained the infrastructure Ostium runs on today, which is a new stack we started building from scratch months before the July attack. When 2026 started, one of our biggest priorities was rebuilding our core infrastructure and tightening OPSEC across the company. AI was progressing so fast that we expected the cost of hacking to go toward zero, and every team, especially in crypto, would have to be paranoid about security to survive. We also needed infrastructure that could support the scale of our ambition. Our original stack got Ostium off the ground, and we always planned to replace it with something bigger and better, the same way we moved our liquidity system step by step toward the hedging model we run today. In March we made the call to build an entirely new stack from scratch, designed both for AI-driven development and for a world of AI-driven attacks. We built it so engineers can run the full test suite locally and see what might break before anything touches production, and so that we can point adversarial models at our own systems to try to take them down. Nobody has standing write access to production and every infrastructure change needs at least another person to approve it. We finished building it in June and started migrating app by app. We went slowly because we were running a live trading system and couldn't risk breaking anything in production. In July, before that migration was done, an attacker compromised an employee's device and used it to steal signing keys from our legacy infrastructure. The timing is brutal because it was exactly the kind of attack the new stack would have prevented. When we halted trading we didn't patch the old system. We moved everything onto the new stack we had already built. We no longer had to protect live trading during the move, so we were able to finish the migration in days and restarted trading 8 days after the attack. Every piece of Ostium now runs on infrastructure we started rebuilding from the ground up in March. I don't think that's widely understood and it's the most important thing in this article. Attacks have only accelerated since July. Binance's report this month describes the same entry point, a compromised engineer's laptop, after which the attackers moved to the people who held deploy rights. They are one of many companies, larger and smaller than us, that have been hit over the last few months. I can’t blame anyone for being targeted; it only shows how dangerous standard practices (e.g. standing write access) have rapidly become in this new world. This article is part 1; parts 2 and 3 will go deeper on the infrastructure behind V3 and the security controls we now run across the company.

kaledora avatar
kaledora@kaledora·Nov 6

We've begun publishing a technical series covering the main design features in Gateway (Ostium V3). This will be followed by a security series. We started with one of the core design principles behind Gateway, offchain proposal with onchain validation. Our first article covered how the new intent-based execution system still enables full validation by the smart contracts. Under the design, orders are not broadcast publicly while they await identification by a keeper and go through a multi-step transaction system, but rather signed and sent privately to the Gateway sequencer before being verified onchain. This significantly improves the latency, privacy, and pricing granularity the system can support. This second article covers how margining and liquidations in Gateway also follow the same principle. The design’s goal is a feature-complete system – cross-margin, aggregated positions, subaccount support, and more – while keeping all the core logic onchain. Margining and liquidations follow a clear set of rules encoded in Gateway’s smart contracts, which any user will be able to verify. The liquidator, for instance, can only propose liquidations; it cannot execute them unilaterally without the public contract logic approving them. We describe these rules and more in this article. Coming articles will cover further features of the new system: cross-asset class margining, settlement and holding costs, the pricing engine, and much more. If you have feedback, questions or thoughts on the design, we'd love to hear from you.

kaledora avatar
kaledora@kaledora·Oct 22

We shared an update today. Adding two thoughts on the structure of Stage 2 and rebuilding revenue: 1. Recovery Structure We designed Stage 1 to reduce the number of affected holders immediately and recover as many people as possible fully upfront. This helps us now operationally focus on a way smaller number of people to recover, and who are our priority going forward. So far, Stage 1 has fully recovered 789 wallets. Stage 2 goes live on October 30th and has both a default liquid state (Category A) and an elective equity option at the close of the next priced round (Category B). We designed Category A to produce a simple catch-all, pro-rata structure that can support a variety of liquid contributions (recovered funds, revenue share, prospective partner contributions, and other tokens). It also maximizes flexibility for OLP loss holders, because it does not require holders to speculate and enables incremental liquidity. Holders can redeem their pro-rata share of funds in the Recovery Contract at any time, as funds come in. We designed Category B for eligible holders who prefer an option with upside potential beyond the value of their losses, enabling them to convert a portion of their then-outstanding losses into an equity stake in Ostium Labs. Recovering affected OLP holders continues to be our number one priority. The split between A and B exists to enable holders to prioritize what they prefer in amounts they determine: cash or participation in potential longer-term upside. Splitting these options will lead to a distribution of resources over time that better reflects OLP loss holder preference. For those who do not wish to elect equity, every dollar converted into equity at the next priced round by other OLP loss holders still decreases the total amount of outstanding losses that would otherwise be covered by liquid sources (e.g. revenue, recovered funds). This means that the same cash covers a larger share of the remaining losses, allowing those losses to be recovered more rapidly, all else equal. 2. Rebuilding revenue The elephant in the room: how do we bring back revenue? Even before the exploit, we were actively pivoting strategy. A big focus of our previous user acquisition was growing the absolute user count as much as possible. That was the wrong approach. 90% of protocol revenue came from the cohort of users who drove significant volume and most valued what differentiates our product: liquidity in size. That is Ostium’s ideal customer profile, and we will go full force on acquiring these users going forward. That means a stronger focus on upmarket, rather than downmarket. Enterprises, smaller institutions, and protail are the best fit for our product. They care about execution, predictable holding costs, access to underlying market liquidity, and the ability to one-shot large trades. This is precisely where Ostium’s edge lies. Gateway was designed with exactly that in mind. It is a significant upgrade to the existing Ostium trading architecture. For the first time, we are embedding into the trading engine cross-margining, aggregated positions, enhanced security limits, an enterprise-grade API, and a privacy-first design. These design decisions are informed by our conversations with early design partners and will continue to be as Gateway nears production. We are focusing on this strategy because we believe it objectively has the highest probability of success. We believe it will work and won’t stop until it does.

marlon avatar
⟳ kaledora retweeted
marlon@momilio·Oct 16

Both Kaledora and Marco took calls with all big OLP holders The convo I had with them was a very transparent from their end, def won back my trust to some degree They also shared their plans for how to get revenue back, and I'm cautiously bullish. Didn't expect it at all, but changed my mind on a lot of things and I'm no longer writing off a comeback

Ostium avatar
⟳ kaledora retweeted
Ostium@Ostium·Oct 13

We've updated the nomenclature and added figures to the Ostium dashboard on Dune to make it clearer how revenue is defined and how it's split between the OLP Vault and the Ostium Protocol. Top-Line Revenue (previously “Total Fees”) is the sum of Trading Fees and Flow Revenue: 󠁯•󠁏󠁏 Trading Fees are the opening fees on each trade, currently split 30/70 between the OLP Vault and the Ostium Protocol. They are charged when the trade is opened. 󠁯•󠁏󠁏 Flow Revenue is spreads, rollover and liquidation fees. Since the liquidity upgrade in April, Flow Revenue goes to the Ostium Protocol. You can see this evolution in the monthly chart. Unlike Trading Fees, Flow Revenue is built into the trade itself (spreads on entry and exit, rollover while the position is open, liquidation fees if it is liquidated) and is settled through the trader’s PnL. Neither value represents gross profits: Top-Line Revenue is an onchain, gross figure and does not include the off-chain cost of hedging flow. Gross profit is an off-chain metric and is the value of what remains after the OLP Vault’s share and off-chain hedging costs. We will continue to make improvements to the Dune dashboard and welcome feedback on how it can be further improved.

kaledora avatar
kaledora@kaledora·Oct 9

Our first update on Stage 2 recovery, which will include: - 50% of Ostium opening fees. - The option to elect an equity stake in Ostium Labs at the next priced round in the same equity class as the team. We will continue to share more updates as the mechanics and percentages are finalized, but are committing to interim updates as we work through the plan. Thank you to the dozens of OLP holders who have taken the time to speak to us over the last weeks.

kaledora avatar
kaledora@kaledora·Oct 7

A brief update on the recovery process: - 708 allocations have been redeemed for a total of 183k USDC. - The number of outstanding OLP holders above the 1,000 USDC threshold not in the Convenience Allocation is 293, down from 345 a week ago. - We will be sharing an update on the design of Stage 2 later this week, informed by our discussions with OLP holders over the last week. Thank you very much to the dozens of OLP holders who have engaged with us and contributed their ideas.

kaledora avatar
kaledora@kaledora·Oct 3

I promised to share a bit more context on Ostium’s finances and revenue model yesterday, as I believe much of the discussion on X is grounded in a misunderstanding of how the protocol monetizes, its costs, and a resulting inflated view of the company’s finances. I will cover both revenues and expenses. 1. Revenue The ~$44.7M top-line revenue number on Dune (linked below) reflects the aggregate value flowing over 2 years across four protocol sources: opening fees ($20.39M), price impact/spreads ($16.5M), rollover fees ($4.75M) and liquidation fees ($3.35M). Since 2024, that top-line revenue has flowed: - $22.14M (49.6%) to the OLP vault: 30% of opening fees (after June 16, 2025; 50% of opening fees before that) plus, until the April 28th 2026 liquidity upgrade, all spreads, rollover and liquidation fees. Because OLP was the direct counterparty to traders, those flows didn’t guarantee OLP returns were always net positive: revenue capture on OLP was sometimes enough and other times insufficient to make up for the times when residual trader deltas worked against the pool. - $19.55M (43.8%) in liquidity costs: $16.05M in hedging and $3.49M in interest on hedging capital. - $2.96M (6.6%) retained as gross profits. Top-line revenue is measured onchain, while take-home gross profits are dependent on off-chain hedging costs that definitionally aren’t reflected on a blockchain. Spreads as a category are computed as the difference between mid- and execution price. Off-chain hedges face spread costs too - no trade executes at mid - as reflected in the liquidity costs category. The same is true for rollover fees; any margined position has holding costs. Ostium is a growth business, and our goal was expansion over short-term monetization. Prior to April, Ostium did not have the ability to monetize flow; the April upgrade was the first step in that direction, and the system had been live for a matter of months when the exploit happened. Finally, we will add more detail and clearer nomenclature to the Dune dashboard over the coming days, including splitting the pre- and post-liquidity upgrade revenue for clarity. We will share a post with details when this has been done. 2. Expenses Ostium raised $300k in 2022, $3.6M in 2023, $4.1M in 2024, and $19.4M in 2025. We raised our Series A into a rapidly growing and highly competitive market. After years of waiting for the market to catch up, it finally had. Now our mandate was to grow, fast. Ostium expanded from ~12 team members in 2025 to 35 in Q2 of 2026, driven first by a wave of engineering hires, and later by an expansion of the growth and marketing teams. Beyond salaries, we invested heavily in data licenses and direct-to-consumer marketing, while bearing heavy cloud infrastructure, auditing, and AI costs. We expanded spend aggressively. In a winner-take-all – or most – market, we made the bet most companies at our stage make: invest everything in growth. In hindsight, we did not build a balance sheet that gave us room to respond to an event like this. We struggled to accept that reality and were silent for far longer than we should have been. That was a mistake. 3. The path forward Growing the business and giving the 345 remaining OLP holders a stake in that growth is the most viable path to recovery we see. We will be sharing more details on Stage 2 recovery in the coming week. The plan will center on shared upside. We have received tremendous input from many of you over the last 48H, and will continue to iterate with OLP holders to design a structure that ties outcomes to the protocol’s success, while being compatible with future financing. Thank you to the many of you who have sent supportive messages, both privately and publicly, over the last days. We recognize it will take time to build back confidence. We will continue fighting, and would like all of you to be on that journey with us.

Ariel Givner avatar
⟳ kaledora retweeted
Ariel Givner@GivnerAriel·Oct 1

After spending months trying to make someone whole that lost their life savings from Drift (only to get completely ignored), I get very *quick triggered* and judgmental when it comes to recovery plans. It broke my heart, and my spirit to a level of which I can’t really describe in words. Deleting my quoted post from this morning and pretending I said nothing would be easy, but it wouldn’t be right. I apologize to Ostium as I reacted too quickly to their recovery plan, I should have kept my mouth shut and dug into it more. I hope that users can be made whole, and despite the initial silence, it’s encouraging to see a team try to make things right. I hope I was wrong, and I apologize for jumping the gun. I apologize to @contrarianmarco and the Ostium team. At the end of the day, we all want people made whole. Hacks are horrible, and I feel for victims.

McKenna avatar
⟳ kaledora retweeted
McKenna@Crypto_McKenna·Oct 1

Ostium is a great team and the situation is unfortunately out of their control. I too have been hacked by state level actors last year and I wouldn't wish it on anyone. In fact to put it bluntly it's one of the most stressful events I have ever lived through and I've been through some shit. The team was early and right. They were discussing equity perpetuals long before any other team I met. My mindset post the state actor hack was you have an obligation to come back and fight your way out of it. I guarantee this is where they are mentally. Do wish the team all the best and hope they can fight their way back.