Deriv built specialised AI agents to defend at scale. Larry, the Security Analyst, triages threats in real-time. Larry, the Security Architect, designs secure systems from stakeholder requirements. Both are live in production, and the research is now open on GitHub.
Deriv Tech
Financial Services
Triq Dun Karm, Birkirkara, Malta 3,172 followers
A hub for curious minds and a diving board for new ideas from our tech team.
About us
We’re a team of curious builders tackling large-scale, low-latency systems, event-driven architecture, and AI-first automation. With 26 years in trading and billions in monthly volume, we move fast, experiment freely, and solve real problems at scale. 𝗪𝗵𝗮𝘁 𝘄𝗲 𝗯𝘂𝗶𝗹𝗱: - Scalable, real-time platforms processing thousands of requests per second - Cloud-native microservices - AI-agentic solutions supporting risk, trading, security, and anti-fraud - Open source tools and emerging tech - Infrastructure for 300+ assets and counting 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 𝘁𝗵𝗲 𝗳𝘂𝘁𝘂𝗿𝗲 𝗼𝗳 𝗳𝗶𝗻𝘁𝗲𝗰𝗵 We believe in sharing the real story behind our work: the trade-offs, the iterations, and the people who make it happen. This blog is where we unpack the tough problems we’re solving, the tech decisions we debate, and the lessons we learn along the way, including the ones that came from things breaking. Whether you're here to learn, contribute, or explore how we work — welcome.
- Website
-
https://deriv.com/derivtech
External link for Deriv Tech
- Industry
- Financial Services
- Company size
- 1,001-5,000 employees
- Headquarters
- Triq Dun Karm, Birkirkara, Malta
- Founded
- 1999
Updates
-
Shipping production AI on a single server where every deploy causes downtime is a problem that only gets worse as you scale. The Applied AI team knew it and rebuilt it from the ground up. Every deployment meant SSHing in, pulling code, bringing Docker down and back up. While it restarted, the service was offline. One crash and the entire stack went down with it. Daniel Lim, Jun Yi Lim, and Yang En Ling walked through the full story at Monash University Malaysia in their session, "No Proof of Concepts: What It Takes to Ship Production AI." The rebuild meant Terraform for infrastructure as code, ECS Fargate for containerised deploys and AWS Inspector scanning as a hard gate before anything ships. Push to master, and it deploys with zero downtime. Swipe through to see what changed. 👉
-
What if every pull request got a security review as thorough as your best engineer, at any scale? At Deriv, our Security Tech Lead, Vishal, built exactly that. Managing 700+ repositories and hundreds of pull requests every week, manual security reviews simply couldn't keep up. Watch below to see his solution 👇
-
The security landscape is shifting fast. Vulnerabilities have spiked, and AI is why. Dhiraj Mishra, Deriv's Security Manager, breaks down what that means in practice and how Deriv is staying ahead.
-
The traditional video production process has nine handoffs, takes up to six weeks and costs thousands of dollars per clip. That's one clip. Across a full campaign calendar, the queue never clears. That's what the Applied AI team set out to solve. Daniel Lim, Jun Yi Lim, and Yang En Ling presented their solution at Monash University Malaysia in their session, "No Proof of Concepts: What It Takes to Ship Production AI." Video-Gen takes a text prompt and returns a branded MP4. The pipeline handles everything in between. Swipe through to see it in action. 👉
-
Deriv Tech reposted this
𝗠𝗼𝘀𝘁 𝗔𝗜 𝗰𝗼𝗻𝘃𝗲𝗿𝘀𝗮𝘁𝗶𝗼𝗻𝘀 𝘀𝘁𝗼𝗽 𝗮𝘁 𝘁𝗵𝗲 𝗱𝗲𝗺𝗼. 𝗧𝗵𝗶𝘀 𝗼𝗻𝗲 𝗴𝗼𝗲𝘀 𝗳𝘂𝗿𝘁𝗵𝗲𝗿. 🎙️ We're LIVE today with lablab.ai's Pawel and Deriv's Dhiraj, talking AI security, enterprise systems, and building production-grade AI that's actually safe.
The AI advantage, secured
www.linkedin.com
-
𝗪𝗲’𝗿𝗲 𝟮𝟰 𝗵𝗼𝘂𝗿𝘀 𝗮𝘄𝗮𝘆 𝗳𝗿𝗼𝗺 𝗼𝘂𝗿 𝗹𝗶𝘃𝗲 𝘀𝗲𝘀𝘀𝗶𝗼𝗻 𝗼𝗻 𝘀𝗲𝗰𝘂𝗿𝗶𝗻𝗴 𝘁𝗵𝗲 𝗔𝗜 𝗮𝗱𝘃𝗮𝗻𝘁𝗮𝗴𝗲. AI and security teams should be finishing each other’s sentences. Otherwise, the gaps become a risk. Join Deriv's Security Manager Dhiraj and Pawel, Co-Founder of lablab.ai, tomorrow at 2:00 PM GST for a practical look at real breaches and actionable defence strategies. 📹 Live on the Deriv LinkedIn page!
-
-
At Deriv, affiliate analytics once meant business intelligence requests, engineering tickets, static dashboards, and waiting for the right data to reach the right team. Amy Manager changed that. This AI agent lives in Slack and acts as an analyst, engineer, strategist, and trainer in one workflow. She answers complex multi-country queries, builds live dashboards in under two minutes, interprets performance risk, and helps teams move from static reporting to real-time business intelligence. Read Mahantesh Ambi's article to discover how Deriv built Amy Manager, our AI-powered business intelligence agent for on-demand partner analytics.
-
-
Stopping the scroll is the most important job an ad has. The Deriv AI team built Gener8, our in-house AI image engine, to make that possible at scale without sacrificing brand quality. At Monash University Malaysia, AI engineers Daniel Lim, Jun Yi Lim, and Yang En Ling shared how we took Gener8 from prototype to production in their session, “No Proof of Concepts: What It Takes to Ship Production AI.” The first outputs weren’t production-ready: spelling mistakes, off-brand fonts, and logos that were almost right. We turned every failure into a rule, with screen masking for pixel-perfect compositing, structured validation at every stage, and quality gates that reject bad outputs before they reach anyone. Swipe through to see the makeover. 👉
-
One week to go! 🙌 If you're building with AI or securing systems that run it, this one's for you. Dhiraj (Security Manager, Deriv) and Pawel (Co-Founder, lablab.ai · CEO, Surge.xyz) are going live to talk about the gap between AI teams and security teams and exactly how to close it. Mark your calendar. You won't want to miss this. 📅 Thursday, May 21st · 2:00 PM GST · Live on Deriv page!
-