Build resilient, production-grade systems across AI, distributed backends, and cloud-native infrastructure — the kind that survive real load, real users, and real chaos.

LLM Evaluation_{Benchmarking systems for reasoning, tool-use, and multi-step AI execution}

Distributed Backends_{High-throughput, low-latency architectures built to scale horizontally}

Automation Pipelines_{Throughput-optimized workflows for document, data, and inference pipelines}

AI Observability_{Tracing, metrics, and evaluation pipelines for models running in production}

Reliability Engineering_{Fault-tolerance, graceful degradation, and SLO-driven system design}

Cloud-Native Deploy_{Container-first, IaC-driven deployments on AWS, Azure, and GCP}

Intelligent Workflow Orchestration_{Multi-agent systems and orchestration layers that coordinate models, tools, and services under real-world constraints}

AI SystemsmeetDistributed Infrastructure

_{Where model latency budgets collide with cross-region replication}

ScalabilitymeetsReliability

_{Where 10× traffic meets a 99.99% availability contract}

Intelligent WorkflowsmeetProduction Constraints

_{Where agent autonomy meets rate limits, budgets, and blast radius}

₹410 Cr national-scale safety infrastructure

Backend and network-oriented workflows for Indian Railways' Train Collision Avoidance System — where a single failure is measured in lives, not logs.

Benchmarking AI at scale

Large-scale benchmarking workflows for reasoning and multi-step AI execution — the systems that decide whether a model is actually good.

High-throughput document intelligence

Document intelligence platform with automated structured extraction — turning unstructured chaos into clean data at scale.

Built from bare metal up

Operating system featuring scheduling, memory management, and filesystem components — written in C, booted in QEMU.

ML forecasting in production

ML-powered forecasting platform with APIs and visualization pipelines — because models are useless until deployed.

Real-time event management

Full-stack event management platform with real-time registration workflows.

Code must be clean before it is clever.Great engineering balances simplicity, scalability, resilience.

Systems should fail safely, recover gracefully, and scale efficiently.Reliability is a feature, not an afterthought.

Build infrastructure for tomorrow — not just demos for today.If others are dreaming of a seat, I'm designing the rocket.

I don't just write code.I engineer systems that shape the future.