Rounak Singh

Software Engineer

Bengaluru, India 85 · Excellent

About

Full-Stack and Backend Engineer who builds systems at the level where performance actually gets decided — instrumenting the Linux kernel directly with eBPF, building a custom database replication engine in Go, and shipping full-stack products with real ML behind them. Comfortable owning a feature from kernel-level telemetry or a database engine up through the API and the UI someone actually uses. Looking for a Software Engineer role building distributed, production-grade systems.

Skills & Expertise (27)

React.js Advanced

8.0/10

Years Exp

JavaScript Advanced

8.0/10

Years Exp

Postgresql Advanced

7.5/10

Years Exp

Go Advanced

7.5/10

Years Exp

Next.js Advanced

7.5/10

Years Exp

Node.js Advanced

7.5/10

Years Exp

Typescript Advanced

7.5/10

Years Exp

Python Advanced

7.5/10

Years Exp

FastAPI Advanced

7.5/10

Years Exp

SQL Advanced

7.0/10

Years Exp

GitHub Actions Advanced

7.0/10

Years Exp

Docker Advanced

7.0/10

Years Exp

PyTorch Advanced

7.0/10

Years Exp

REST APIs Advanced

7.0/10

Years Exp

Express Advanced

7.0/10

Years Exp

Distributed Systems Advanced

7.0/10

Years Exp

Tailwind CSS Intermediate

6.5/10

Years Exp

InfluxDB Intermediate

6.5/10

Years Exp

Grafana Intermediate

6.5/10

Years Exp

MongoDb Intermediate

6.5/10

Years Exp

Redis Intermediate

6.5/10

Years Exp

Java Intermediate

6.5/10

Years Exp

Isolation Forest Intermediate

6.0/10

Years Exp

RAG Intermediate

6.0/10

Years Exp

Vector Search Intermediate

6.0/10

Years Exp

C++ Intermediate

6.0/10

Years Exp

LangGraph Intermediate

5.5/10

Years Exp

Work Experience

Developer

Kernel-Native AIOps

Present - Present

Captured kernel-level scheduling, I/O, memory, and lock-contention signals invisible to standard APM tools, by instrumenting five Linux kernel tracepoints (sched_switch, sys_enter/exit, page_fault_user, do_futex) with eBPF at under 1% CPU overhead. Enabled remote, firewall-free telemetry streaming from the instrumented host to the backend, by deploying the eBPF collector on AWS EC2 and tunneling live metrics through ngrok into a FastAPI/InfluxDB pipeline. Cut incident root-cause diagnosis from minutes to under 5 seconds with zero hallucinated findings, by combining a tri-model anomaly-and-causality engine (Isolation Forest, Granger causality, trend regression) with a ReAct LLM agent restricted to structured tool calls.

Developer

StudyGenie

Present - Present

Architected a full-stack learning platform end to end — React frontend, Express API gateway, MongoDB data layer, and JWT auth — to handle real user uploads, sessions, and study history. Kept the UI responsive under heavy document uploads, by building an async pipeline (Cloudflare R2 presigned uploads feeding BullMQ/Redis background workers) that processes documents off the main request thread. Decoupled the ML and RAG layers from the core web app, by serving a PyTorch model and retrieval pipeline behind a separate FastAPI service with clean internal APIs, so either could be redeployed independently.

Developer

ViewSync

Present - Present

Built an Incremental View Maintenance engine in Go that keeps PostgreSQL materialized views updated in real time, cutting update time from over 12 seconds to about 3 milliseconds compared to a standard scheduled refresh. Implemented Change Data Capture by streaming Write-Ahead Log events through PostgreSQL's logical replication protocol, removing the need for table locks or repeated full-table scans. Designed a concurrent in-memory state store with primary and secondary indexes to resolve multi-table joins instantly, applying every update through atomic upserts to keep the data consistent.