I'm a final year CSE undergraduate at LNMIIT and a Founding Engineer at Flexzistay, where I build and scale backend systems in production.

I enjoy working on systems where the engineering gets interesting under the hood: distributed coordination, concurrency, networking, storage, low-latency execution, and performance.

Lately, I've been exploring AI infrastructure, LLM inference, autonomous coding agents, and cloud-native systems. I also contribute to open-source projects in the Kubernetes ecosystem.

I learn best by building things from scratch, measuring how they perform, and understanding why they work.

Repository · TypeScript, LangGraph, Node.js, OpenAI API

LLM-powered code review agent supporting CLI and GitHub App workflows.

• Built multi-stage reviews with parallel diff analysis, line-level grounding, and secondary verification to reduce false positives.
• Added repository-specific policies, token budgets, and evaluation using precision, recall, and F1.

Repository · Go, Raft, WAL, Docker

Kafka-inspired distributed message broker built from scratch.

• Achieved 18K messages/sec with 200 concurrent producers.
• Implemented Raft consensus, leader election, persistent storage, replication, and consumer groups.

Repository · C++20, FIX 4.4, Async I/O

Low-latency trading gateway for order-management and execution workflows.

• Built asynchronous networking and lock-free queues, benchmarking 250K orders/sec.
• Achieved 350µs P50 and 633µs P99 latency.

Repository · Python, FastAPI, vLLM, Prometheus, Grafana

OpenAI-compatible gateway for routing inference requests across vLLM workers.

• Implemented load-aware routing, health checks, automatic failover, retries, and backpressure.
• Built observability and benchmarking for TTFT, token throughput, and request latency.

Live Website · Node.js, TypeScript, PostgreSQL, Redis, GCP

Built the production backend for a hotel marketplace serving 10K+ users.

• Shipped 100+ REST APIs and integrated hotel inventory and booking workflows across 10K+ hotels.
• Implemented concurrency-safe bookings and reduced API P95 latency by 50% to under 150ms.

SIP, Streaming Audio, STT/LLM/TTS, MCP

Built backend infrastructure for real-time voice agents.

• Reduced perceived audio pipeline latency from 1.4s to 900ms.
• Implemented provider-agnostic telephony, call orchestration, failover, and MCP tool execution.

Contributing to Kubernetes-native model serving and cloud-native AI infrastructure.

• PR #142 — Namespace-scoped resource filtering
  Implemented namespace-aware resource filtering using environment variables, backend API changes, and Kubernetes RBAC to support multi-tenant deployments.

• PR #161 — InferenceGraph support
  Added InferenceGraph custom-resource support, including schema integration, backend APIs, and UI resource lifecycle management for multi-model inference pipelines.

I enjoy contributing to projects where improvements in APIs, resource management, and infrastructure make the developer experience better for everyone.

• Competitive programming: LeetCode Knight, contest rating 1856, ranked in the top 6.13% globally.
• Problem solving: 500+ coding problems solved across platforms.
• Education: B.Tech in Computer Science and Engineering, LNMIIT.

I'm always interested in backend engineering, distributed systems, AI infrastructure, and meaningful open-source work.

Feel free to reach out if you're building something interesting in these areas.