π Quick Overview β click to toggle
- π B.Tech Computer Science @ Graphic Era Deemed University (graduating May 2027, GPA 8.25/10)
- π Seeking SWE / AI-ML / Data Science internships
- π» Full-stack developer β Python, TypeScript, C++
- π§ Passionate about LLM applications, geospatial ML, RAG systems
- ποΈ Production experience: FastAPI, Next.js, PostgreSQL, AWS / OCI
- ποΈ AWS Certified Cloud Practitioner + OCI 2025 AI Foundations Associate
- π Top 20% on LeetCode (200+ solved) Β· Top 15 / 900+ teams at Graph-E-thon 3.0
- π οΈ Technical Programs Lead @ The Designovation Hub
π² Fun fact (click me)
I once optimized a delivery-route solver using bitmask DP (O(2βΏΒ·nΒ²)) β turns out the fastest way to visit every node is also the fastest way to burn a CPU fan.
π§ Expand to see breakdown
π Proficiency breakdown (click to expand)
Python ββββββββββββββββββ 90%
TypeScript ββββββββββββββββββ 75%
C++ ββββββββββββββββββ 70%
SQL ββββββββββββββββββ 70%
| Framework | Proficiency | Projects |
|---|---|---|
| FastAPI | βββββ | promptgit, Face Recognition HRMS, AeroGraphFed |
| Next.js | ββββ | Face Recognition HRMS |
| Streamlit | ββββ | promptgit, DocChat |
| LangChain | ββββ | DocChat, RAG systems |
π Click to expand stats
π Live contribution snake (auto-generates once the workflow below is added to this repo)
πΌ Click a project to expand details
π Interview-oriented prompt versioning & A/B testing platform
Stack: PostgreSQL Β· FastAPI Β· Streamlit Β· Redis Β· Anthropic SDK
β¨ Features
- Git-like version control for prompts (fork, branch, merge)
- RBAC + Redis-backed caching for <50ms prompt retrieval
- Queue-driven metrics workers, prompt A/B testing & analytics
- GitHub Actions CI Β· 48 tests
π View Repository β
π€ HR management with real-time face recognition for attendance & identity
Stack: FastAPI Β· Next.js Β· pgvector Β· Docker Β· PostgreSQL Β· OpenCV
β¨ Features
- Real-time face detection & recognition (99.2% accuracy)
- Automated attendance tracking with geolocation
- Employee database with facial embeddings via pgvector similarity search
- Advanced modules: CCTV integration, email notifications
- <200ms inference latency per frame
π View Repository β
π°οΈ End-to-end satellite imagery processing for environmental analysis
Stack: XGBoost Β· GeoPandas Β· FastAPI Β· Rasterio Β· Docker
β¨ Features
- ETL pipelines for satellite & geospatial data (GeoTIFF, Shapefile)
- Vegetation index (NDVI) & terrain feature engineering
- XGBoost land-use classification served via REST API
- 10x speedup via vectorization Β· 80% test coverage Β· GitHub Actions CI/CD
π View Repository β
π Privacy-first document Q&A with local LLMs & multi-hop reasoning
Stack: Ollama Β· LangChain Β· LangGraph Β· ChromaDB
β¨ Features
- 100% local β no data sent to external APIs
- Multi-hop reasoning with LangGraph for complex queries
- Semantic search over PDF, TXT, Markdown, and web content
- Streaming responses, <300ms typical response time
π View Repository β
π¬ What I'm currently exploring
- π€ Advanced LLM techniques: fine-tuning, LoRA, quantization
- π Distributed systems: message queues, event streaming (Kafka)
- π Time-series ML: forecasting, anomaly detection
- ποΈ System design: scalable architectures, microservices
- π― Production ML: model monitoring, CI/CD for ML pipelines

