B.Tech IT @ IIIT Bhopal ('27) β’ Core Contributor to Django, FastAPI, Hugging Face & LangChain
I actively architect, optimize, and maintain production features in leading open-source repositories:
| Organization & Repo | Core Work & Technical Impact | PR / Issue | Status |
|---|---|---|---|
Django Coredjango/django |
Engineered & merged reusable qualname() utility into django.utils.module_loading, removing duplicate dotted-path logic across 4+ submodules (tasks, migrations, mail). Collaborated with Django Fellows. |
#21875 | π£ Merged in Core (100% Coverage / 2K+ Tests) |
LangChainlangchain-ai/open-swe |
Refactored modular prompt-section tables and cleaned deprecated constants in autonomous software engineering agent pipelines. | #2416 | π£ Merged |
Hugging Facehuggingface/transformers |
2Γ Merged Contributor β Resolved model instantiation class bugs in SigLIP2 architecture and fixed Python syntax compilation runtime errors in VibeVoice pipelines. | #48197 #48489 |
π£ 2x Merged |
Microsoftmicrosoft/PyRIT |
Implemented Garak Divergence red-teaming scenario, custom DivergenceScorer, 42 prompt seeds dataset, and 12 unit tests for GenAI risk benchmarking. |
#2547 | π’ Active / CLA Signed |
n8nn8n-io/n8n |
Implemented TLS Server Name (SNI) support for Microsoft SQL credentials & connection pooling in TypeScript. | #37683 | π’ Approved (5/5 Review Score) |
FastAPItiangolo/fastapi |
Offloaded blocking synchronous I/O to async threadpools (reduced benchmark latency by 20%β50%); fixed recursive typing for `Json[list[T]] | None`. | #16213 #16220 |
LamaticLamatic/Lamatic |
Engineered automated bug-to-test-case synthesis engine merged into main; built Notion Assistant workflow integration. | #318 #387 |
π£ Merged / π’ Clean |
EvalPortadhabnr-ux/evalport |
Developed production evaluation dataset adapters for Parea AI and Humanloop LLM evaluation platforms. | #19 #26 |
π£ Merged (2 PRs) |
| System | Architecture & Stack | Key Metrics / Benchmarks | Links |
|---|---|---|---|
| RAG Document Assistant | FastAPI, Groq LLaMA-3, FAISS, Docker | <700ms latency, 99%+ uptime session retrieval | Code β’ Demo |
| Emotion Recognition API | PyTorch, Custom CNN, FastAPI, Docker | 82% Accuracy (FER-2013), <150ms inference | Code β’ Demo |
| E-Commerce Microservices | Django REST, Redis, Sentence-Transformers | 30% latency reduction with semantic search caching | Code β’ Demo |
| LLM Guard & Refusal Engine | Python, PyTest, Docker, Safety Evaluator | Dual-mode safety evaluation with 20+ unit tests | Code β’ Demo |
Enterprise Backend API built with Java 21, Spring Boot 3.3.5, and Spring Security 6.
- Session-based authentication, transaction CRUD, savings goals, and financial analytics engine.
- 100% test pass rate (86/86 unit & integration tests) with >86% code coverage.
Production platform for benchmarking AI Agents and GenAI applications.
- Automated RAG retrieval quality measurement, hallucination detection, token cost tracking, and latency analytics.
Scalable SaaS backend built with Node.js, Express, PostgreSQL, and Docker.
- Features tenant data isolation, Stripe subscription billing webhooks, and cryptographic API key management.
AI Safety & Alignment research tool built in Python.
- Detects psychological manipulation tactics in prompts and reformulates them into ethical requests.
- Languages:
PythonJavaC++JavaScriptTypeScriptSQLBash - Backend & Systems:
FastAPIDjangoFlaskNode.jsExpressRedis (Caching & Queues)CeleryMicroservicesRESTful APIs - AI / ML & GenAI:
PyTorchHugging Face (transformers)LangChainLlamaIndexFAISSRAG ArchitecturesSentence-TransformersGroq LPU - Databases & Storage:
PostgreSQLMySQLRedisVector Databases - DevOps, Cloud & Tools:
Docker & Docker ComposeLinux / ShellGitHub Actions (CI/CD)PostmanRenderVercel

