package main
type Engineer struct {
Name string
Role string
Company string
Depth string
Focus []string
}
func main() {
me := Engineer{
Name: "Sarthak Shyngle",
Role: "Senior SRE / Platform Engineer",
Company: "Alpheya",
Depth: "kprobe → cgroup → pod → cluster",
Focus: []string{"Go", "eBPF", "Kubernetes", "GitOps", "Observability"},
}
me.MakeSystemsBoring() // reliability is a feature
}- 🔭 Now: Senior Site Reliability / Platform Engineer at Alpheya — scaling from deep-tech niches into large production systems.
- 🧪 Before: Founding Engineer & Senior SWE at Keploy — turning real API traffic into test suites + mocks, with zero code changes to the app under test.
- 🐧 Deep end: eBPF (kprobes, cgroup socket programs), L4/L7 transparent proxies, TLS interception, Linux internals.
- ☸️ Kubernetes: mutating webhooks, custom controllers & CRD testing, Helm + FluxCD GitOps.
- 🎓 Google Summer of Code mentor, 3 years running (2023–25) — see below.
- 🏆 #3 all-time contributor at Keploy OSS · 300+ PRs reviewed.
- 💬 Ask me about: DSA, Go concurrency, API testing, kernel-space networking.
- ⚡ Off-keyboard: sports and any excuse for a bit of adventure.
Every summer since 2023 I've mentored GSoC contributors through systems programming and Go internals — the parts of open source that scare people off. Proposal review, architecture guidance, code review, and the long unglamorous middle where a contributor turns into a maintainer.
|
Systems programming eBPF, syscalls, network proxies, kernel-space debugging |
Go internals Concurrency, memory model, profiling, runtime behaviour |
Review culture 300+ PRs reviewed; turning first-time contributors into maintainers |
Selected mentorship — GSoC 2024
| Project | Contributor | Co-mentors |
|---|---|---|
| Cross-Platform Extensions: Keploy IDE Integration | Akash Singh | Animesh Pathak, Gourav Kumar |
Considering applying to GSoC with Keploy? Proposals in the systems/eBPF space are the ones I most like to see.
🔍 Expand for the full breakdown
Languages
Systems & Networking
Platform & Delivery
Data & Storage
| Project | What it does | Stack |
|---|---|---|
| 🧬 Keploy — Core Platform | Standalone binary → distributed microservices over IPC sockets. 2× faster, 85% quicker test runs | Go IPC |
| ☸️ K8s Proxy & CRD Testing | Mutating webhook injects agents into live pods for in-cluster record & replay | Go controller-runtime |
| 🐝 eBPF Interception Engine | kprobes + cgroup socket programs → codeless mocks via an L4/L5 transparent proxy | eBPF C Go |
| 🔥 Profiler-MCP | MCP server letting LLMs analyze perf bottlenecks and auto-launch flamegraphs | MCP Go |
| 🤖 AI Unit-Test Generator | VS Code extension (10k+ installs) + LLM PR agent with Tree-sitter AST parsing | LLMs TypeScript |
| 🐘 Postgres Wire-Protocol Parser | Binary protocol parser tuned for 2k+ ops/sec live capture and replay | Go Performance |
🔍 Expand full project breakdowns
|
Led redesign from a standalone binary to distributed microservices, isolating GC-heavy workloads into separate processes over IPC sockets — 2× performance gain. Built a centralized mock-caching layer that cut test execution time by 85%.
|
Designed the HLD and led development of Keploy's K8s proxy — a mutating webhook injecting agents into live pods for in-cluster traffic record & replay. Captured and replayed cluster traffic to test custom controllers / CRDs.
|
|
End-to-end interception using kprobes + cgroup socket programs, enabling codeless mock generation through an L4/L5 transparent proxy. Dynamic CA certificate injection to intercept TLS traffic transparently.
|
An MCP server letting LLMs (Claude, Copilot) autonomously analyze performance bottlenecks across Go, Python, and Java from runtime profiling data — auto-launching interactive flamegraphs.
|
|
VS Code extension (10k+ downloads) plus an AI-powered GitHub PR agent using LLMs, Tree-sitter AST parsing, and vector-based semantic retrieval.
|
High-performance parser for the PostgreSQL wire protocol, tuned to handle 2k+ ops/sec for live capture and replay.
|
📜 More: earlier work
- Java Agent (ByteBuddy) — runtime bytecode instrumentation to generate E2E tests inside an Apache Karaf environment.
- Java SDK with middleware filters — captured every request/response of any Spring Boot app as tests + mocks, with zero code changes.
- gRPC server — protobuf-based contract enabling multi-language SDK integration without redefining models per language.
- Integration testing pipelines — templated generation + deduplication cut 10k+ test YAML templating from 5 min → 0.5 s; socket-based agent tracked per-request line coverage via JaCoCo.
- FarEye (Data Engineering Intern) — large-scale ETL on Databricks / PySpark / Hadoop, geocoding optimized with Uber H3; automated unit-test pipelines for Redshift and DataLake snapshot validation.
- 🥉 #3 all-time contributor at Keploy OSS; top contributor and owner of the enterprise edition, maintaining repo health and reviewing 300+ PRs.
- 📊 Org impact — led client delivery and issue resolution, stabilizing production testing pipelines.
- ✍️ Technical writing — blogs on engineering internals and automated testing:
| Post | Topic |
|---|---|
| Maintaining Auto-Generative API Tests: the Need to De-duplicate | Test maintenance at scale |
| Exploring Unit Test Generative Tools | AI-assisted testing |
| What is the Postgres Wire Protocol | Binary protocols |
| How to Test Traffic with a Custom Kubernetes Controller | K8s controllers & CRDs |
| Integration Testing vs End-to-End Testing | Testing strategy |
→ All posts on the Keploy blog
Auto-updated daily from Hacker News, TechCrunch AI, and The Verge — see workflow
- [OpenAI says it slowed Astra model development over security concerns](https://techcrunch.com/2026/08/07/openai-says-it-slowed-astra-model-development-over-security-concerns/) Aug 7- After Rippling blew millions on AI in months, it built an employee ROI tool Aug 7
- Fenix Flexin isn’t even denying using AI to make ‘Rubberz’ anymore Aug 7
- Watching Roku’s AI channel is like eating from a trough Aug 7
- OpenAI puts the brakes on a new model because it’s supposedly too powerful Aug 7
- Assembly Hall of Shame Aug 7
Large-scale system design · SRE practice at scale · Advanced observability · Chaos engineering
Open to interesting problems in platform engineering, reliability, and developer tooling.