Projects
Building Platforms That Scale
Sentinel
AI-Backed SRE Agent
An autonomous SRE agent that combines advanced AI reasoning patterns with custom-built infrastructure tools, enabling it to plan, execute, and evaluate root cause analysis tasks — reducing incident investigation time and freeing engineers to focus on higher-order problems.
During beta, Sentinel achieved ~60% accuracy in autonomous incident investigation, with plans for production rollout.
- ~60% accuracy in autonomous RCA detection during beta
- Proactive detection of issues before user impact
- Context-aware recommendations using Tree-of-Thought reasoning
Odin
Multi-tenant Deployment Platform (Open Source)
A multi-tenant deployment platform serving both internal engineering teams and external sister companies. Odin handles deployments at IPL-scale peak traffic with zero-downtime and isolated environments per tenant.
Odin follows a Docker-like architecture — just as Docker images can be built in any language and Docker orchestrates them without knowing internals, Odin components implement the Odin Component Interface and the platform orchestrates them seamlessly. Components can be written in any language, decoupling orchestration from implementation.
Now open sourced with 24+ components — MySQL, Kafka, Spark, PostgreSQL, Redis, Cassandra, and more.
- 15M+ concurrent users during IPL season
- 24+ open-sourced components (MySQL, Kafka, Spark, PostgreSQL…)
- Zero-downtime deployments with tenant isolation
Scaler
Auto-scaling platform for Dream11
Real-time auto-scaling platform that eliminated 1-hour pre-planning delays and improved infrastructure utilization.
- Reduced scaling decisions from 60 minutes to under 5 minutes
- Direct revenue increase during live IPL matches
- Optimized cloud costs through intelligent scaling
Optimus
Database-as-a-Service Platform
Multi-tenant DBaaS platform serving 150+ services with automated provisioning, backups, and failover.
- Reduced database provisioning time from days to minutes
- Automated failover with <90 second RTO
- Compliance-ready with encryption at rest and in transit
Earlier Work
CXMonitor
Observability platform that reduced production bugs by 70% through proactive monitoring and automated review.
Crossover
ZALP
Social recruitment platform for procurement professionals with AI-powered candidate matching.
Zycus