# Pranay Mundra

> distributed-systems engineer — fault injection, tail-latency attribution, tracing & observability, SDK design. Based in San Francisco / Santa Cruz, CA. Open to distributed systems, infrastructure, observability and reliability roles. Contact: pmundra@ucsc.edu · https://github.com/pronei · https://www.linkedin.com/in/pranay-mundra-115055189.

This site is a static Hugo build. Full corpus for agents (bio, every project writeup, CV, depth notes): https://pronei.github.io/llms-full.txt. CV as PDF: https://pronei.github.io/cv/pranay-mundra-cv.pdf. Search index: https://pronei.github.io/searchindex.json.

## Experience

### University of California, Santa Cruz — Graduate Teaching Assistant (Jan 2025 – Jun 2026)
- Mentored students in CSE115A/B/C on software projects sponsored by Nutanix and Keysight Technologies — requirements analysis, high-level system design, sprint planning.

### MakeMyTrip & Goibibo (NASDAQ: MMYT) — Software Engineer I (Jun 2022 – Jun 2024)
- Part of India's largest online travel agency, on the platforms behind A/B experimentation, discounting and persuasion.
- Designed a highly concurrent multi-org ConfigKey store for the A/B platform — two-level caching on Aerospike, sub-millisecond p99 at 300k requests/second, down from 50ms.
- Built a rule-based rate limiter optimizing notification delivery across SMS, WhatsApp and email — more relevant engagement, up to 40% lower infra cost.
- Revamped loyalty programs ([MMTBLACK](https://tech.makemytrip.com/introducing-the-all-new-mmtblack-indias-ultimate-travel-loyalty-program-that-keeps-on-giving-83a20ebee850), [GoTribe](https://www.goibibo.com/loyalty/), [myPartner Hero](https://mypartner.makemytrip.com/agent/loyalty/hero/?crId=0d2930d8-532a-4a16-be35-a26b44ff23f1)) with configurable per-user aggregations, enabling new program types for any customer segment.
- Closed a fraud hole in Refer&Earn (~$50K) by constraining mobile-number reuse in the signup flow.
- Added virtual-thread support to the legacy Java 8 A/B platform with quasar, cutting CPU footprint up to 40%.

### MakeMyTrip & Goibibo — Software Intern (Jan 2022 – May 2022)
- Set up cache sync for intra-cluster services using Kafka; parallelized /assignVariant on the A/B platform.
- Supported the legacy route-planning service ahead of its [How2Go](https://www.makemytrip.com/how2go) rework.

### ARTPARK @ Indian Institute of Science — Research Intern (Jun 2021 – Dec 2021)
- Built a log aggregator on Elastic alerting warehouse robot operators to environmental anomalies.
- Designed a ROS2 safety node integrated with warehouse cameras to stop robots on human proximity.

## Education
- University of California, Santa Cruz — Master of Science in Computer Science (GPA 3.89), Sept 2024 – Aug 2026. Advanced Computer Architecture, Distributed Systems, Advanced OS, Algorithms.
- PES University, Bengaluru — B.Tech in Electronics & Communication (CGPA 9.01), Aug 2018 – Aug 2022. Control System Theory, Digital Communication, Robotics; TA for [UE19EC339](https://github.com/pronei/UE19EC339-Robotics).

## Research and working style
- Research thread — faults-lab, a fault-injection platform that isolates each microservice's contribution to tail latency under load; three public Go repos at github.com/microfaults — atropos (boundary-level fault-injection SDK with a cache-box replay primitive), zeus-go (load orchestration with k6 sidecars and a Vegeta policy engine), manteion (experiment oracle that schedules campaigns and correlates W3C Baggage traces)
- Builds MCP servers — slug-mcp (Rust) exposes UCSC campus services and Santa Cruz environmental telemetry as tool calls (public connector at https://2262-cse115b-02.be.ucsc.edu/mcp); also an Adobe Premiere Pro MCP server
- Languages — Go and Rust by preference, Java when the platform demands it, Python, C/C++, SQL, TypeScript/JavaScript, Shell; instruments everything with OpenTelemetry
- Names systems after classical mythology and the names are load-bearing — atropos cuts the thread, zeus throws the lightning, manteion is the oracle at Delphi

## Projects
- [faults-lab](https://pronei.github.io/projects/faults-lab/): A fault-injection research platform — inject faults at service boundaries, isolate each service's contribution to tail latency under concurrent load. — Go, OpenTelemetry, gRPC, Kubernetes, k6, Vegeta, Postgres — source: https://github.com/orgs/microfaults/repositories
- [global upgrade safety & MSS](https://pronei.github.io/projects/upgrade-safety/): Static analysis that decides whether simultaneous schema changes across a microservice mesh can co-deploy safely — and computes the maximal safe subset when they can't. — Go, OpenAPI, Protobuf — source: https://github.com/pronei/gus-mss-poc
- [slug-mcp](https://pronei.github.io/projects/slug-mcp/): An MCP server for banana slugs — UCSC campus services and Santa Cruz environmental telemetry as structured tool calls for LLMs. — Rust, rmcp, Moka cache — source: https://github.com/pronei/slug-mcp
- [chatcut 3.0](https://pronei.github.io/projects/chatcut/): Natural-language video editing on real footage — a Premiere Pro plugin and a standalone desktop editor sharing one AI backend. Vibe editing, not AI-generated video. — Python, FastAPI, TypeScript, React, Tauri, Rust, Premiere Pro UXP — source: https://github.com/pronei/ChatCut-3.0
- [nogo](https://pronei.github.io/projects/nogo/): A lightweight, rules-driven rate limiter — the open-source descendant of the notification gate I built at MakeMyTrip. — Go — source: https://github.com/pronei/nogo

## Currently (live from the GitHub API at build time, 2026-10-28)
- manteion (experiment oracle): operational — docs: add Manteion oracle banner to README (pushed 2026-09-09T01:28:21Z), ci success
- atropos (fault-injection SDK): operational — docs: add Atropos banner to README; align net… (pushed 2026-07-14T19:06:37Z), ci success
- zeus-go (load orchestration): operational — docs: add Zeus banner and a Status & roadmap … (pushed 2026-07-29T00:32:45Z), ci success
- chatcut 3.0 (natural-language video editing): operational — Added video to video generation mode with swi… (pushed 2026-09-08T19:44:35Z)
- slug-mcp (campus MCP server): operational — docs: mark the auth-gated tools as non-functi… (pushed 2026-09-30T22:36:53Z), ci success

## Skills
- languages: Go, Rust, Java, Python, C/C++, SQL, JavaScript, Shell
- frameworks: OpenTelemetry, gRPC, vert.X, Spring, gin, k6, Vegeta
- devops: Docker, Kubernetes, Jaeger, Prometheus, Grafana, Jenkins, AWS, git
- data & queues: MongoDB, MySQL, Aerospike, Redis, Kafka, RabbitMQ
- expertise: distributed systems, fault injection, tracing & observability, SDK design, scalable web services

## Pages
- [about](https://pronei.github.io/about/): Who I am, how I got here, and why everything is named after Greek mythology.
- [cv](https://pronei.github.io/cv/): Education, experience, and skills — rendered from data, with the real PDF one click away.
- [contact](https://pronei.github.io/contact/): Email is fastest. No forms, nothing to track you.