Bengaluru SRE Meetup at Smarsh

We’re excited to host a community meetup hosted at the Smarsh office this May.
This meetup is focused on topics like Agentic AI, Site Reliability Engineering and Observability, with an emphasis on practical experiences and real-world learnings from engineers working close to production systems.
This meetup is intended for engineers working in or interested in SRE, DevOps, platform engineering, observability, and cloud operations. Whether you’re running systems at scale or just getting started, the goal is to keep the conversations practical, honest, and useful.

When: May 23rd, 2026
Location: Smarsh, Bengaluru
Address : Second Floor, 9, Cambridge Rd, Halasuru, Udani Layout, Bengaluru, Karnataka 560008

Agenda:
10:00 AM - 10:25 AM -
Introductions and Kick -off

10:30 - 11 AM -
Talk 1 : AI in the space of observability
Before going fully into “agentic-workflow” how can we integrate the core AI constructs like Skills, Workflow, Workpacks into our Observability space today. The idea is to use these tools from Day1 of development cycle. The talk explores the idea of where / how to use these constructs in Observabilitiy.

Speaker: Akshay Deshpande (Smarsh)

11:10 - 11:40 AM -
Talk 2 - Scaling Tempral workers with KEDA and VPA
In this talk, we'll look at :

• Temporal primer — what it is, the core architecture, and where workers fit in

• Temporal vs Argo — quick comparison of the two, what's different about the execution model, and why that matters for how you scale workers

• Worker slots — the concept that's central to how scaling decisions get made

• The scaling problem — what worker scaling looks like in practice, and why horizontal alone isn't enough

• Why both KEDA and VPA — how we landed on combining them, and the failure modes when either is missing or misconfigured

Speaker: Sanil Khurana (Atlan)

11:50 - 12:20 PM -
Talk 3 - Architecting Oodle Logs Engine on S3 and Lambda
This talk is about architectural blueprint of how the team at Oodle designed their logs engine for cost efficiency and performance in comparison with Clickhouse.
Many modern observability products are built on ClickHouse (for example, Dash0 and HyperDX/ClickStack). ClickHouse is a strong analytical database, but it was designed before S3-first storage and serverless compute were practical defaults. Oodle was designed around those primitives from the start. Oodle achieves 10x faster queries and 3-8x lower cost than ClickHouse-based competitors in production workloads.
The core architectural difference is simple: ClickHouse keeps storage and compute in one always-on cluster, while Oodle separates them. Data stays in S3, and compute appears only when a query runs. The rest of the trade-offs follow from that. This talk will go over architectural blueprint of Oodle's Logs engine and its comparison with ClickHouse.

Speaker: Gaurav Maheshwari (Oodle AI)

12:20 PM onwards -
Networkig over Coffee, Refreshments