Understory ranks No. 140 on Inc. 5000 list of fastest-growing companies in U.S. Read more about Understory on the 2026 Inc. 5000 list

Paddy Srinivasan

Paddy Srinivasan

Chief Executive Officer at DigitalOcean

Greater Boston · Developer tools · LinkedIn profile

Paddy Srinivasan is Chief Executive Officer at DigitalOcean, based in Greater Boston. On LinkedIn Paddy writes mostly about company news and industry analysis, most often as text posts.

8.2K+followers
324reactions on a typical post
241reactions on the best post
2 of 12recent posts worth featuring

Paddy's best LinkedIn posts

Recent posts that drew 100 reactions or more and say something a peer would learn from, most reactions first. Click any post to open it on LinkedIn.

Paddy Srinivasan
Paddy SrinivasanChief Executive Officer at DigitalOcean
In the transition from experimentation to full-scale production, AI workloads expose real architectural challenges: throughput bottlenecks, latency SLAs, and cost economics. Meeting these at scale requires more than “more GPUs” — it demands hardware-software co-design, systems-level tuning, and production-grade inference orchestration. A compelling example is the collaboration between Character.ai, DigitalOcean, and AMD announced today: 🔹 Workload scale: Character.ai operates one of the most demanding inference workloads — serving ~20M users and over a billion requests per day, with high concurrency and strict latency targets. 🔹 Performance gains: By deploying AMD Instinct MI300X/MI325X GPUs on DigitalOcean’s unified inference platform and optimizing runtimes (ROCm + vLLM + AITER), Character.ai achieved ~2× higher production throughput under the same latency envelope versus generic GPU infrastructure. 🔹 Economic impact: Beyond performance, the solution delivered ~50% lower cost per token and expanded usable capacity, materially improving TCO and operational efficiency. These results underscore a critical shift in AI infrastructure: 👉 Inference isn’t a commodity: At scale, production AI workloads require tight integration across compute (e.g., Instinct GPUs with 192 GB HBM3), optimized runtimes, scheduler intelligence, and platform reliability. 👉 Platforms matter: DigitalOcean’s Inference Platform shows how an inference-first cloud—treating GPUs as an engineered stack, not interchangeable resources—unlocks meaningful performance and economic gains. Hardware–runtime co-design: Optimized inference on 8× MI325X (HBM3, FP8) nodes using ROCm + AITER + vLLM, enabling high-efficiency FP8 execution paths and eliminating kernel fallbacks common in generic environments. Parallelism topology optimization (DP/TP/EP): For the Qwen3-235B FP8 MoE model, shifting from DP1/TP8/EP8 to DP2/TP4/EP4 reduced inter-GPU communication and improved expert routing, delivering ~2× higher sustained throughput at identical p90 latency. Memory efficiency via FP8 KV cache: FP8 KV cache cut per-request memory footprint by ~50%, enabling higher batch sizes and concurrency without HBM exhaustion—improving tokens/sec/GPU and cost per token. Production-grade orchestration: Deployed on DOKS with NUMA-aware scheduling, topology-aligned device plugins, and prefix/KV reuse, delivering stable, horizontally scalable inference at billions of daily requests under real-world SLAs. https://lnkd.in/eMcx8ai4
Paddy Srinivasan
Paddy SrinivasanChief Executive Officer at DigitalOcean
DigitalOcean CEO Paddy Srinivasan told me revenue growth broke 40% in the third quarter, the fifth straight quarter it beat on key metrics, and the company raised its outlook to 44% to 45% growth for the fourth quarter and 60%-plus for 2027. He credited software as agentic systems drive token and compute demand; its tokens 10x'd last quarter. He said RPO doesn't drive the business, citing its 680,000-plus customers and little customer concentration. Full Fortt Knox conversation linked in the comments:
Image from Paddy Srinivasan's LinkedIn post

What Paddy posts about

  • Company news83%
  • Industry analysis8%
  • Customers8%

What makes the best ones work

  • Most of Paddy's best posts open with a plain statement of a view.
  • 50% of them are text only, with no image or video.
  • They run long, often past a thousand characters.

Questions about Paddy Srinivasan

Who is Paddy Srinivasan?

Paddy Srinivasan is Chief Executive Officer at DigitalOcean, based in Greater Boston. On LinkedIn Paddy writes mostly about company news and industry analysis, most often as text posts.

What does Paddy Srinivasan post about on LinkedIn?

Mostly company news, industry analysis and customers. Of the 12 recent posts we read, 2 drew 100 reactions or more with real substance.

What is Paddy Srinivasan's most popular recent LinkedIn post?

The one that opens “In the transition from experimentation to full-scale production, AI workloads expose real architectural challenges: throughput bottlenecks,…”, with 241 reactions and 4 comments.

Every post belongs to Paddy Srinivasan. Understory is not affiliated with Paddy Srinivasan or DigitalOcean. We read public LinkedIn posts to study what B2B leaders write and what their audiences respond to, and the bio is our own summary. If this is you and you would like the page changed or removed, email alex@understoryagency.com and we will do it.

More developer tools thought leaders

See every developer tools thought leader, or the ads DigitalOcean runs.

Want posts like these on your profile

We ghostwrite LinkedIn posts for B2B founders and executives, in their own voice, and turn the audience into pipeline.