Subvolts take Understand AI inference infrastructure and company growth from Baseten's CEO. Nov 18, 2025 - Developers
- Learn about optimizing AI model serving, inference tools, and tuning for new hardware.
- Business
- Understand the infrastructure challenges and company-building strategies in the fast-growing AI inference market.
A solid interview with the CEO of a rapidly growing AI inference company, discussing key technical and business challenges.
The CEO Behind the Fastest-Growing AI Inference Company | Tuhin Srivastava
Audio streams from Lukas Biewald. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
Lukas Biewald interviews Tuhin Srivastava, CEO of Baseten, about the AI inference company's rapid growth. They discuss challenges in serving large models, differences between inference tools like vLLM and TensorRT-LLM, optimizing for new hardware, and scaling a company in a fast-moving market. The conversation covers Baseten's journey and the market's alignment with their infrastructure.
This episode explains how companies like Baseten help AI models run fast and efficiently. It covers the technical hurdles in making big AI systems work smoothly, which is important for anyone using or building AI applications.
This episode offers insights into the practical challenges of AI inference, a critical bottleneck for deploying large models. It's for AI engineers and infrastructure leaders who need to understand optimizing performance and reliability. The discussion on tools and hardware tuning is relevant as frontier labs like Google Gemini and OpenAI push larger, more complex models, requiring efficient deployment solutions.
- HostLukas Biewald
- GuestTuhin SrivastavaCEO and founder, Baseten
- Guest: Tuhin Srivastava, CEO and founder of Baseten
- Show: Gradient Dissent
- Length: 59 minutes
- Topic: AI inference ecosystem
- Tools discussed: vLLM, TensorRT-LLM, SGLang
- Hardware mentioned: B200
In this episode of Gradient Dissent, Lukas Biewald talks with Tuhin Srivastava, CEO and founder of Baseten, one of the fastest-growing companies in the AI inference ecosystem. Tuhin shares the real story behind Baseten’s rise and how the market finally aligned with the infrastructure they’d spent years building.
They get into the core challenges of modern inference, including why dedicated deployments matter, how runtime and infrastructure bottlenecks stack up, and what makes serving large models fundamentally different from smaller ones.
Tuhin also explains how vLLM, TensorRT-LLM, and SGLang differ in practice, what it takes to tune workloads for new chips like the B200, and why reliability becomes harder as systems scale.
The conversation dives into company-building, from killing product lines to avoiding premature scaling while navigating a market that shifts every few weeks.
Connect with us here:
Tuhin Srivastva: https://www.linkedin.com/in/tuhin-srivastava/
Lukas Biewald: https://www.linkedin.com/in/lbiewald/
Weights & Biases: https://www.linkedin.com/company/wandb/
- AI inference
- Tuhin Srivastava
- How AI models process information and generate outputs after being trained.
- vLLM
- Tuhin Srivastava
- An open-source library for fast and efficient large language model inference.
- TensorRT-LLM
- Tuhin Srivastava
- NVIDIA's software library for optimizing large language model inference.
- SGLang
- Tuhin Srivastava
- A Python library for easier and more efficient LLM deployment.
1 h 341 h 15He's Building an AI That Can't Lie | Dan Klein
WithDan Klein
1 h 15He Raised $70M to Cure Every Disease With AI | Samuel G. Rodriques, Edison Scientific
WithSamuel G. Rodriques
Same guests, other shows
43 min
49 minKoray Kavukcuoglu: “This Is How We Are Going to Build AGI”
WithKoray Kavukcuoglu
45 min
1 h 13He Co-Invented the Transformer. Now: Continuous Thought Machines - Llion Jones and Luke Darlow [Sakana AI]
WithLlion Jones, Luke Darlow
Questions people ask
- What are the core challenges of modern AI inference?
- Challenges include dedicated deployments, runtime and infrastructure bottlenecks, and the fundamental differences in serving large versus smaller models.
- How do inference tools like vLLM and TensorRT-LLM differ?
- The episode discusses practical differences between these tools and SGLang for optimizing AI model serving.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Lukas Biewald.