┌─────────────────────────────────────────────────────────────┐
│ GAURAV DUBEY · Senior Engineer │
│ Expedia Group · Gurgaon, India · ↑ since 2022 │
└─────────────────────────────────────────────────────────────┘
I build agentic systems that catch fraud before it ships — LLMs orchestrated through deterministic graphs, grounded in policy, and observable enough to put on call.
11 years in. Java & distributed systems → Python & ML platforms → now full-time on agent infrastructure.
|
A workbench for prompt iteration. Edit a prompt, sweep it across a real dataset, let a judge model score the result, accept the suggested edits, promote the winner. The same engine runs in CI to gate regressions. TypeScript · v2.2.1 · single-file binary, local-first |
Autonomous open-source contribution plugin for Claude Code. Orchestrates six specialist agents — issue selection, repo profiling, planning, building, merge-probability scoring, comment resolution — without hand-holding. Claude Code plugin · v0.4.1 · seven merged releases |
last synced · 2026-07-21
┌─ impact ──────────────────────────────────────────────────────────────┐
│ 49 merged · 17 open · 1.6M⭐ reached
│ ▁ ▅█▁▂▃▁ ▁ ▁▂▂ ▃ ▃▂ prs over last 30 days
└───────────────────────────────────────────────────────────────────────┘
| repository | ⭐ | merged | open | latest title |
|---|---|---|---|---|
| obra/superpowers | 258k | #1987 | — | test: realign antigravity + pi mapping assertions with pruned refer... |
| affaan-m/ECC | 231k | #2429 #2413 #2401 #2399 #2386 #2374 #2373 #2372 #2369 #2339 #2177 #2163 #2161 #2159 #2134 #2041 #1701 #1682 #1669 #1490 | #2491 #2490 #2488 #2466 | fix(opencode): resolve command agent ids to registered opencode agents |
| huggingface/transformers | 162k | #45611 | — | Raise clear error for problem_type="single_label_classification" wi... |
| google-gemini/gemini-cli | 106k | #25822 | #28183 | fix(vscode-ide-companion): preserve terminal focus when closing dif... |
| ant-design/ant-design | 98k | #58241 | #58551 | fix(Typography): inherit font size in editable textarea |
| tesseract-ocr/tesseract | 75k | #4563 | — | Fix crash when LSTM is missing in disabled-legacy build (#4448) |
| unslothai/unsloth | 68k | #7113 #7082 | #7156 | Studio: add option to disable the in-memory API monitor |
| mem0ai/mem0 | 61k | #5202 | — | fix(vector_stores): pass namespace as top-level kwarg to UpstashVec... |
| headroomlabs-ai/headroom | 60k | #1513 | — | fix(proxy): retry passthrough on transient upstream connection close |
| go-gitea/gitea | 56k | #38479 | — | fix(packages): serve noarch Alpine index for any requested architec... |
| run-llama/llama_index | 50k | #21891 | #22371 | refactor(core): deduplicate SimpleDirectoryReader load_file/aload_file |
| ClickHouse/ClickHouse | 48k | #106916 | — | Common: mask sensitive data on Exception::addMessage |
| apache/airflow | 46k | #65685 | — | Honor AUTH_ROLE_PUBLIC in FastAPI API server |
| mitmproxy/mitmproxy | 44k | #8326 | #8294 #8293 #8280 | proxy: bracket IPv6 literals in upstream CONNECT authority |
| danny-avila/LibreChat | 41k | #13171 #12825 | — | 🧹 chore: Type Agent MCP lean projection in ServerConfigsDB |
| conductor-oss/conductor | 32k | #1239 | — | fix(rest): log 4xx client exceptions at WARN instead of ERROR |
| Lightning-AI/pytorch-lightning | 31k | #21686 | #21796 #21791 #21759 | fix: log the convenience epoch metric once per step (#20902) |
| better-auth/better-auth | 29k | #10368 | #10423 #10383 | fix(two-factor): reject reused TOTP codes on step-up verification |
| envoyproxy/envoy | 28k | #45510 | — | docs: remove broken Ubuntu noble install instructions |
| mochajs/mocha | 22k | #6016 | #6120 | feat: print each error within AggregateErrors |
| oxc-project/oxc | 22k | #23909 #23015 #22752 | — | fix(linter/unicorn/prefer-at): skip object numeric-key access |
| lima-vm/lima | 21k | #5165 #5090 | — | shell: clean up stale ssh.sock on --reconnect when the master is al... |
| pydantic/pydantic-ai | 18k | #5815 #5681 | — | Auto-enable Anthropic Files API beta when messages contain Uploaded... |
| langchain4j/langchain4j | 12k | #5272 | — | Skills: add DefaultFileSystemSkill.toBuilder() and round-trip tools... |
Sorted by upstream star count. Closed-unmerged PRs skipped.
- Role — Senior Engineer at Expedia Group, on the Fraud & Risk platform team
- Building — multi-agent fraud reasoning over LangGraph + MCP
- Shipped — privacy-preserving LLM gateway with three-tier tokenisation, eval-gated in CI
- Open to — collaboration on agent infrastructure, eval tooling, MCP servers
AI LangGraph · LangChain · MCP SDK · FastAPI · Pydantic · Presidio · spaCy
Models Bedrock · SageMaker · Claude · OpenAI · Groq · Ollama · OpenRouter
Data PostgreSQL · DynamoDB · Redis · Elasticsearch · Kafka · Airflow
Core Java · Python · Golang · TypeScript · AWS · Docker · K8s · Vault
Crypto HKDF · HMAC-SHA256 · CRC8 · per-conversation key derivation
Eval LangSmith · OpenTelemetry · Grafana · property-based · golden sets
- Benchmarks over vibes. Datastores picked with IOPS and
$/hr; LLM routes picked with eval numbers. - Security is the first tenet, not the last. Name the threats, name the controls.
- Entity thinking. If you're debugging individual events, you're already behind the fraud ring.
- Email — gauravdubey0107@gmail.com
- LinkedIn — linkedin.com/in/gaurav1792
- GitHub — @gaurav0107