Building @ExographDev in @rustlang. Author, AspectJ in Action. Former core team member for #cloudfoundry @VMware and #spring.

San Francisco Bay Area
Joined July 2008
Ramnivas Laddad retweeted
So excited to introduce Infinity, an agent runtime in Rust for extreme concurrency! Infinity brings a new architecture that treats agent concurrency as a scheduling problem, tool call primitives for asynchronous events, and native support for serverless. github.com/hydro-project/inf…
4
8
1
50
2,060
Ramnivas Laddad retweeted
It was awesome to share our work at @awscloud with the Agentic AI community!
Shadaj Laddad from @awscloud is tackling exactly that at AGNTCon + MCPCon NA with “Predictable Correctness for Unpredictable Code.” The goal? Use Hydro and Rust to catch distributed systems errors at compile time, even when agents write the code. Catch the keynote live: bit.ly/4dVkjdO
2
4
26
1,479
Ramnivas Laddad retweeted
Amazing presentation by @ShadajL at AGNTCon!
Shadaj Laddad from @awscloud is tackling exactly that at AGNTCon + MCPCon NA with “Predictable Correctness for Unpredictable Code.” The goal? Use Hydro and Rust to catch distributed systems errors at compile time, even when agents write the code. Catch the keynote live: bit.ly/4dVkjdO
3
16
1,267
Ramnivas Laddad retweeted
Super excited for this! The principles behind Hydro, “universal properties”, are increasingly valuable across all software domains. Will be sharing how we think about the role of type systems and simulation in the agentic era, and where Hydro is going next. Join us!
How do you verify code when agents write faster than humans can review? @ShadajL, Research Lead for Hydro at AWS, has some answers—on the AGNTCon + MCPCon North America keynote stage. Get the details: bit.ly/4xAwe8g Join him October 22-23 in San Jose, CA!
1
1
16
1,715
Switched to GPT/Codex.... let's see how it goes!
1
2
164
This looks amazing. I can imagine using it with coding agents to store/retrieve interesting discoveries.
Introducing @OpenKnowledge, the best markdown IDE for humans and agents. Open source. Local and private. LLM-wiki ready. Use with Claude, Codex, and your favorite agent today.
1
3
1,061
Ramnivas Laddad retweeted
New blog post! shadaj.me/writing/cyberchase… If you've been following the latest Lean + RL research, and are curious about the technology at its core, this is for you! I introduce the basics of Lean in the context of an episode from the PBS Kids show Cyberchase.
3
8
1
36
10,345
This year’s conference was absolutely fantastic! Looking forward to the next one!
Big news: #dev2next is coming back next year in Denver — October 12–15, 2026! 🎉 4 days of cutting-edge sessions, future-focused keynotes, and meaningful collaboration with the minds shaping what’s next in software. 💻 Secure your spot soon! 🚀
3
6
3,956
At ⁦@dev2next⁩ attending ⁦@starbuxman⁩ presentation. So much energy! When I grow up, I want to be a presenter like Josh!
1
1
6
4,560
Exograph now supports MCP! exograph.dev/blog/exograph-n… LLMs need rich data, but you need bulletproof access control. Writing secure SQL queries that handle both is a pain. With Exograph, you define your schema + access rules. LLMs get the data they need, users only see what they should.
186
Ramnivas Laddad retweeted
Learn how the MCP helps LLMs go beyond RAG—powering smarter, real-time interactions while keeping data secure and systems reliable with @ramnivas at dev2next this Fall! 💥 dev2next.com/speaker/b15040f… 🎟️ dev2next.com 👀 Use code JOIN-RAMNIVASL-50OFF for a discount!
1
1
194
I will be presenting a technical deep dive on Model Context Protocol (#MCP) at Silicon Valley JUG this Wednesday, July 2nd. See you there! meetup.com/sv-jug/events/308…
2
1
3
1,681
Ramnivas Laddad retweeted
Another #svgenai meetup is a wrap. 1) 📜 slides are posted here : svgenai.dev/meetups/2025-06-… video recordings coming soon. 2) 👉 Sign up to be notified of future events: lu.ma/svgenai Checkout our website: svgenai.dev/ 3) 💰 I gave out credits for @nebiusaistudio at the meetup. DM me if you'd like to get some credits to try out latest open source models in Nebius AI Studio studio.nebius.com/ 4) 🎤 Big thanks to our speakers - @MelGoesTech on Langflow - @ramnivas about #MCP - I gave an update on Allycat (github.com/The-AI-Alliance/a…) open source project. #meetup #AIMeetup #allycat @davenielsen @thealliance_ai
1
3
194
Ramnivas Laddad retweeted
🚀 SVGenAI June Meetup – MCP & Langflow ⚡ Lightning Talks: - AllyCat Update: ​🎤 Talk 1: Langflow by Melissa Herrera of DataStax (@DataStax) & DevRel Langflow (@langflow_ai) 🎤 Talk 2: Model Context Protocol (MCP) Deep Dive by Ramnivas Laddad (@ramnivas) of Exograph (@ExoGraphDev) 🙏Hosted by @ASAComputersInc @RackLive 🗓️: 530pm , Thursday June 26, 2025 📍: Fremont, CA 🍕 Pizza & beverages provided 👉 sign up: lu.ma/uz10jjfs?tk=6N0vtj #GenAI #Langflow #MCP #SVGenAI #AICommunity #OpenSource #DeveloperMeetup
1
2
197
Ramnivas Laddad retweeted
My PhD dissertation is now live, featuring many chapters of brand new, not-yet-published work! www2.eecs.berkeley.edu/Pubs/…
5
5
53
4,259
Attending ⁦@ShadajL⁩’s PhD dissertation talk. He worked really hard, cared a lot about doing things right, and pushed the boundaries of computer science. I am really proud of my boy!
9
2
43
2,873
I've noticed that most (if not all) MCP (Model Context Protocol) servers written in TypeScript/JavaScript using `npx` as the JavaScript engine. That's... a lot of trust. `npx` gives full access to local system resources, unrestricted network calls, and environment variables--which feels like leaving all your doors and windows open because it's convenient. A far more secure default would be Deno (@deno_land), which flips the model: 🚫 No access to file system, network, or env by default ✅ You explicitly opt-in via flags like `--allow-read=/home/temp` or `--allow-net=api.example.com` This kind of permission model drastically reduces the attack surface--especially critical in the context of tool poisoning attacks (invariantlabs.ai/blog/mcp-se…), or just the general risk of a locally running server going rogue. With Deno, developers can allow access only when it makes sense for that tool to do its job. Of course, Docker can achieve similar objective--but network restrictions aren't nearly as straightforward to configure. WebAssembly is another promising option, but very few servers target it today. Given Deno's "no trust by default" approach, I'm surprised that it isn't the go-to choice.
3
2
196
@rough__sea, this feels like a perfect use case to highlight the power of @deno_land’s "secure by default" model.
93