Software Development Engineer 2 @ Microsoft | Ex- Amazon, Ex- Razorpay
Distributed Systems • Cloud Infrastructure • High-Throughput Ingestion • Search Infrastructure • Cost Engineering

🌐 shashankswe2020-ux.github.io/shashankswe2020-ux — interactive portfolio
_{Built with Anime.js v4 · Barba.js · Three.js (WebGPU) · GSAP · Theatre.js — source in /site}

I design and operate distributed systems that sustain high-throughput workloads under unpredictable traffic while maintaining availability and cost efficiency.

Currently working as an SDE 2 (L62) at Microsoft, building scalable cloud infrastructure and services. Previously part of the indexing charter for Amazon OpenSearch Serverless (AWS) — building infrastructure powering vector search and large-scale search workloads.

I focus on:

• Throughput-aware scaling
• Backpressure & flow control
• Observability-first architecture
• Failure-mode resilience
• Cost predictability at scale

• Building scalable cloud infrastructure and services at Microsoft
• Applying distributed systems expertise to large-scale product engineering
• Driving system design, performance optimization, and reliability improvements

Scalability & Control

• Designed and implemented internal Index Rollover API enabling ingestion auto-scaling
• Built controlled backpressure mechanism → reduced availability drops by 90%
• Led 5TB+ ingestion benchmarking to identify system bottlenecks

Observability & Reliability

• Designed telemetry for ingestion, shard recovery, flush & refresh metrics
• Built dashboards for vector search indices (LLM embedding workloads)
• Improved signal-to-noise ratio in operational alerts

Business & Customer Impact

• Resolved 100+ high-severity performance escalations
• Onboarded major enterprise customer → +$2M/month revenue
• Mitigated cost anomaly → $8.7M reduction in customer billing

• Designed backend for high-throughput tax payment workflows → ₹6.4 Crore launch month revenue
• Owned payout-links microservice as SME
• Fixed state inconsistency bug preventing ₹4 Crore financial exposure

Core

• Java
• Golang
• C#
• .NET
• Distributed system design
• AWS & Azure infrastructure

Secondary

• C++
• Python
• Docker
• Kubernetes
• GraphQL

• Flow control algorithms in ingestion-heavy systems
• Vector indexing for large-scale embedding workloads
• Cost-aware distributed architecture
• Lucene internals & indexing pipelines
• Benchmarking & load simulation frameworks

MCP server to connect to the WHOOP API. Built with TypeScript, it enables seamless integration with WHOOP's health and fitness data through the Model Context Protocol. Use can add it as connector in claude desktop too

Staff-Engineer-in-a-Box: curated, BM25-searchable backend & distributed-systems intelligence across 20 domains and 12 language stacks — drop it into Claude Code, Cursor, Copilot, or any AI assistant.

AI-powered research paper agent that orchestrates spec-driven development, planning, code review, documentation, and shipping — all through a multi-agent architecture built for end-to-end research workflows.

• CodeChef — 2050 (5★)
• Codeforces — 1618 (Expert)
• Google Code Jam Round 2
• Facebook Hacker Cup Round 1

Strong algorithmic foundation complements production systems work.

  B.E., Birla Institute of Technology & Science, Pilani (2020)

Design for failure.
Measure before optimizing.
Scale deliberately — not accidentally.

Impact-driven distributed systems engineering.

A real, animated multi-page portfolio site lives under /site and deploys to GitHub Pages on every push to main.

Run locally:

cd site
npm install
npm run dev

See site/README.md for build & deployment details.