Rootconf 2026 Annual Conference

Rootconf 2026 Annual Conference

Platforms for AI, and AI for Platforms

Tickets

Loading…

Platforms for AI, and AI for platforms

20–21 November 2026 · TERI Auditorium, Domlur, Bengaluru · Online via livestream

New conference dates: 20–21 November

Rootconf 2026 Annual Conference has moved to 20–21 November, as many community members will still be away during the earlier dates. The venue remains TERI Auditorium, with livestream access for remote participants.

Alosh Denny’s quantization workshop will take place on 22 November. The chaos engineering workshop remains on 28 November. Four further talks and additional workshops will take place at a members-only event on 12 December.

AI is changing the platforms we build, and the platforms themselves are now a constraint on what AI can do in production.

At Rootconf 2026, meet engineers working through these challenges: serving models, reviewing a flood of AI-generated code, handling agent-generated queries and keeping stateful systems reliable.

This is a two-day, practitioner-led conference for people who build, run and make decisions about platforms. Expect production case studies, technical demonstrations and discussions about what teams tried, what failed and what they would change.

Get your Rootconf membership and attend → https://hasgeek.com/rootconf/2026/#tickets

View schedule here: https://hasgeek.com/rootconf/2026/schedule

Rootconf audience


Spotlight session: controlling the code flood

Inside PhonePe’s code-review platform, Arnold

Tushar Naik · Software Architect, PhonePe

AI can multiply the volume of code an organisation produces. It does not multiply the number of careful human reviewers—or remove the need for them. So where does the bottleneck go?

Tushar will unpack how PhonePe built and scaled Arnold: an asynchronous, distributed code-review platform that orchestrates micro-agents while keeping humans in the loop. This production case study covers the build-versus-buy decision, the architecture behind the review pipeline, the limits of a simple AI loop and the adoption work needed to make an internal platform useful.

Watch the demo and examine what makes code review effective, when an in-house platform is justified and where AI should assist rather than decide.

Read the full submission → https://hasgeek.com/rootconf/2026/sub/how-phonepes-in-house-code-review-platform-arnold-NuJAg6r8TWdQnEqekev872


More November programme highlights

When ML serving becomes a distributed systems problem

Vikash Kumar · Software Engineer, InMobi

Running hundreds of models reliably turns model serving into a control-plane problem. Vikash follows a production model deployment through readiness, scaling, version rollouts, recovery and retirement—and examines how a Kubernetes CRD becomes a long-lived platform API.

Read the submission →

48 GPU hours, 10% of the data: debugging an LLM training run

Alosh Denny · Aleddo Technologies/Intensors

A training job consumed 48 GPU hours while repeatedly processing a tenth of its intended dataset. Alosh traces the debugging journey through shared-state failures, missing checkpoints and the checks worth running before committing more compute.

Read the submission →

Surviving the query-pocalypse!

Gaurav Maheshwari · Founding Engineer & India Site Lead, Oodle AI

AI agents explore query shapes, work around errors and may keep retrying inefficient requests. Gaurav examines how this changes monitoring, error messages, performance signals and query routing for a production observability platform.

Read the submission →

The complete programme and session timings will be published with the schedule.


Put the ideas into practice: November workshops

Rootconf members can register for these hands-on workshops, with in-person participation and hybrid access for remote participants.

22 November · Break a model, then fix it

A hands-on quantisation workshop with Alosh Denny

Shrink an open model to fit the hardware you have—and prove it still works. Compare memory, speed and capability at different bit widths, and build an evaluation that catches silent failures a headline benchmark can miss.

For platform, infrastructure, MLOps, SRE and serving engineers working with a fixed compute budget.

Explore the quantization workshop → https://hasgeek.com/rootconf/quantization-workshop/

28 November · Chaos engineering for stateful applications on Kubernetes

Test failover, recovery and data durability in PostgreSQL on Kubernetes.

Use a three-instance CloudNativePG cluster, Chaos Mesh, Prometheus, Grafana and Jepsen to test the guarantees behind “automated failover.” Examine failover time, replication-lag recovery, write-loss windows and what happens when the control plane is unavailable.

Explore the chaos engineering workshop → https://hasgeek.com/rootconf/chaos-engineering-workshop/


12 December · A members-only day of talks and workshops

Continue the conversation with four sessions on production agents, infrastructure safety, recovery and observability, alongside additional workshops.

poster

The following talks will take place on 12 December. This event is open to Rootconf members only.

Hermes in production: building an AI infrastructure engineer that we could actually trust

Saurabh Hirani, Principal SRE at One2N
How do you let an infrastructure agent do useful work without giving it unrestricted production access? Saurabh shares Hermes’s production architecture: Slack as the interface, read-only investigation tools, Vault-backed secrets and proposed changes through GitOps pull requests. The central design choice: let the agent investigate, reason and propose, while keeping execution deterministic and auditable.

Read the submission →

Beyond SSE: building resumable agent streams

Aditya Singh · Staff Engineer, Glance (InMobi)
An agent can outlive its client connection—and the client may reconnect through a different gateway. Aditya explains how Glance separates execution from delivery, using Redis Streams for ordered, per-turn replay. Explore how defining the recovery contract shapes the state model, technology choices and operational trade-offs.

Read the submission →

TraceQL internals: building a query language for distributed traces

Suraj Nath · Senior Software Engineer, Grafana Labs
How do you query petabytes of deeply nested trace data on object storage? Suraj explores Grafana Tempo’s columnar storage, query planning and sharding, and the trace-specific questions behind TraceQL’s design. A look at the trade-offs between storage cost, indexing and query latency.

Read the submission →

From AI agents to AI platform: lessons from building an in-house multi-agent platform

Manas Chaturvedi · Director–Senior Technical Architect, IDfy
Moving agents into production makes orchestration, state, recovery, cost and evaluation shared platform concerns. Manas explains IDfy’s move from standalone agents to an internal multi-agent platform, prioritising predictable execution over autonomy. He also examines how evaluations and A/B testing help validate improvements before promoting new versions.

Read the submission →

The December schedule, participation details and additional workshops will be announced separately.


One membership, a year of Rootconf

Your membership gives you access to all Rootconf activities for one year, including:

  • The Annual Conference on 20–21 November
  • The November workshops
  • The members-only event on 12 December
  • Topical Edition on Databases in June 2027 in Bengaluru and online

Keep comparing approaches with practitioners as new engineering questions arise throughout the year.

Team bookings receive 5% off for 5 or more tickets, and 10% off for 10 or more tickets.

Book one year’s access to Rootconf → https://hasgeek.com/rootconf/2026/#tickets


The questions behind the programme

Platforms for AI: Inference, model serving, agent workloads and the infrastructure, observability, security, governance and cost controls they require.

AI for platforms: AI applied to incident response, debugging, deployments, security workflows, code review and developer tooling.

Reliability, databases, distributed systems and day-to-day operations remain central. If you are a platform engineer, SRE, DevOps engineer, security engineer, MLOps engineer, data-infrastructure practitioner or engineering leader, bring your questions and experience.


Meet the companies supporting Rootconf

Meet engineers from Nutanix, our Platinum Sponsor, to learn about their work on resilient platform systems, discuss their products and explore relevant career opportunities.

You will also meet teams from Roku Inc (Silver Sponsor), Fastah (Bronze Sponsor) and ClawMetry (Community Sponsor), and learn about the systems and tools they are building.


Reviewed by practitioners, for practitioners

Community editors review proposals, discuss them with speakers, review rehearsals and help strengthen the evidence, trade-offs and practical lessons in each session.

  • Bharadwaj Embar - Engineering Manager, Platform Engineering, Nutanix
  • Rolland Santimano - Principal Engineer, Rubrik
  • Talina Shrotiya - Senior Engineering Manager, Couchbase
  • Vivek Pemavat - Principal Engineer, Acceldata

“Unlike the traditional ‘submit and wait’ approach, RootConf gives speakers an opportunity to demo their talk before a group of reviewers and receive genuine, constructive feedback on what to improve, add, or refine.”

“The audience was technically strong, curious, and genuinely eager to learn.”

Aftab Mehrab
Presenter, Rootconf 2025 Annual Conference
Production Engineer – Infrastructure, Comply

Aftab Mehrab at Rootconf


Volunteer with Rootconf

Help shape and run Rootconf 2026. Contribute to registration and on-ground coordination, editorial and workshop coordination, live notes, session summaries, or design for stickers, swag and the venue.

Apply to volunteer →

Rootconf community


Participate

20–21 November 2026
TERI Auditorium, Domlur, Bengaluru · Online via livestream

Get your Rootconf tickets → https://hasgeek.com/rootconf/2026/#tickets

For inquiries, email info@hasgeek.com or call/text +91 7676332020.

Featured submissions

See all

Venue

The Energy Resources Institute (TERI)

4th Main Road, Domlur II Stage,

153, 2nd Cross Rd, Phase 3, BDA Colony, Stage 2, Domlur

Bengaluru - 560071

Karnataka, IN

Loading…

Hosted by

We care about site reliability, cloud costs, security and data privacy

Supported by

Platinum Sponsor

Nutanix is a global leader in cloud software, offering organizations a single platform for running apps and data across clouds.

Silver sponsor

Engineers at Roku Bengaluru, building the tech behind the #1 TV streaming platform in the US and Mexico. Come say hello at our meetups.

Bronze Sponsor

Internet infrastructure APIs for IP geolocation and more

Community Sponsor

Real-time Observability & Governance layer for AI agents