Building AI systems, scalable backends, and intelligent applications.
🎓 B.Tech student passionate about Software Engineering & Artificial Intelligence
🤖 Building with LLMs, Transformers, RAG & AI Agents
⚙️ Developing backend systems and APIs with Python & FastAPI
🧠 Solved 300+ DSA problems across multiple coding platforms
🌎 Exploring and contributing to Open Source
🚀 Learning primarily by building real systems
Python · Java · C · JavaScript · TypeScript
HTML · CSS · SQL
FastAPI · REST APIs · WebSockets
PostgreSQL · SQLAlchemy · SQL
Redis · Firebase · Supabase
JWT · OAuth · Authentication · Authorization
Docker · OpenAPI · Swagger
Transformers · Large Language Models
Tokenization · Embeddings
Hugging Face · Hugging Face Models
Model Inference
Fine-Tuning · Transfer Learning
Model Adaptation · Pre-trained Models
RAG · Agentic RAG
Vector Search · Hybrid Search
Embeddings · PGVector
Retrieval Pipelines
AI Agents · Agentic Systems
Tool Calling · Agent Memory
LangChain · LangGraph
Multi-Agent Systems
Vision Models · Multimodal LLMs
Image Understanding
TensorFlow · LightGBM
Scikit-learn · Pandas · NumPy
Feature Engineering
Classification · Fraud Detection
React · JavaScript · TypeScript
Tailwind CSS · HTML · CSS
Git · GitHub · VS Code
Docker · Kubernetes
Netlify · Storybook
A hybrid AI + deterministic financial decision system that determines whether a requested expense can be safely afforded.
- 🧠 AI-based financial information interpretation
- 👁️ Multimodal amount extraction
- 🏦 Financial state reconstruction
- 📈 90-day cash-flow forecasting
- 💳 Payment strategy generation
- 🧮 Deterministic financial simulation
- 🛡️ Financial safety validation
- 📊 Payment-plan evaluation
- ⚡ Deterministic candidate ranking
- ✅ Final decision validation
Financial Evidence
│
▼
┌───────────────────┐
│ AI Perception │
└─────────┬─────────┘
▼
Structured Data
│
▼
Financial State
Reconstruction
│
▼
90-Day Simulation
│
▼
Payment Strategies
│
▼
Safety Validation
│
▼
Deterministic Ranking
│
▼
Validated Decision
Stack
Python · LLMs · Gemini · OpenRouter
Multimodal AI · Financial Simulation
A machine-learning system focused on detecting suspicious and coordinated fraudulent behavior from financial transaction data.
- 📊 Feature engineering
- ⚖️ Imbalanced classification
- 🌳 Gradient boosting
- 🔬 Model experimentation
- 📈 ROC-AUC evaluation
- 🎯 PR-AUC evaluation
- 🔍 Fraud pattern analysis
- 🧪 Model comparison
Stack
Python · LightGBM · Scikit-learn
Pandas · NumPy · Machine Learning
A backend system designed around financial transactions, concurrency, consistency, security, and reliable state management.
- PostgreSQL row-level locking with
SELECT ... FOR UPDATE - Handles concurrent fund transfers safely
- Prevents race conditions and double-spending
- Locks sender/receiver account rows during transfers
- Atomic multi-step fund transfers
- Explicit transaction boundaries
- Commit / rollback handling
- Maintains consistency across account, transaction, and ledger updates
- Records debit and credit operations
- Links ledger entries to transactions and accounts
- Maintains an auditable financial history
- Rolls back together with the transaction when failures occur
- Redis-backed idempotency keys
- Prevents duplicate financial operations
- Stores completed transaction responses with TTL
- Retry requests return the previously processed result
- Redis-based request tracking
- Sliding-window rate limiting
- Uses
WATCH / MULTI / EXECfor atomic updates - Protects sensitive transaction endpoints
- JWT authentication
- Access + refresh tokens
- Redis-backed sessions
- Protected endpoints
- Token validation and expiration
- Explicit rollback on failed operations
- Handles expected application errors separately
- Prevents partial financial state
- Safe failure responses for unexpected errors
- Records important financial actions
- Associates actions with users and transactions
- Stores action descriptions and context
- Provides traceability for financial operations
- Relational data modeling
- Foreign-key relationships
- Unique constraints
- SQLAlchemy ORM
- Database-level consistency
- Alembic migrations
- Version-controlled database schema
- Reproducible database changes
API / Controllers
│
▼
Services
│
▼
Repositories
│
▼
PostgreSQL
- Controllers handle HTTP concerns
- Services contain business logic
- Repositories handle database operations
- Transaction boundaries controlled by the service layer
- Transaction history APIs
- Pagination
- Structured response schemas
- Repository-based querying
Redis
├── Idempotency
├── Sessions
├── Rate Limiting
└── Cached Responses
- OpenAPI specification
- Swagger UI
- ReDoc
- Pydantic request / response validation
- GitHub Actions
- Automated testing workflows
- Automated deployment workflows
Stack
Python · FastAPI · PostgreSQL
SQLAlchemy · Redis · Alembic
JWT · Docker · GitHub Actions
I've solved 300+ DSA problems across multiple coding platforms, focusing on pattern recognition, optimization, complexity analysis, and problem solving.
Arrays
Strings
Hashing
Two Pointers
Sliding Window
Binary Search
Sorting
Stacks
Monotonic Stack
Linked Lists
Trees
Graphs
BFS
DFS
Heaps
Greedy
Dynamic Programming
Understand the Problem
↓
Identify the Pattern
↓
Design the Approach
↓
Analyze Complexity
↓
Implement
↓
Optimize
↓
Test Edge Cases
I enjoy working with real-world codebases, understanding existing systems, fixing problems, and contributing improvements back to the community.
🔍 Find a Problem
↓
🐛 Reproduce It
↓
🧠 Understand the Codebase
↓
🛠️ Implement a Solution
↓
🧪 Test
↓
📦 Open Pull Request
↓
🌎 Contribute Back
AI/ML · LLMs · Developer Tools
Python · Backend · Open Source AI
🔗 Explore My GitHub Contributions →
Python
│
▼
Backend Engineering
│
▼
Machine Learning Basics
│
▼
Transformers & Hugging Face
│
▼
LLMs
│
▼
RAG
│
▼
AI Agents
│
▼
Agentic RAG
│
▼
Multi-Agent Systems
│
▼
Production AI Systems
- 🧠 Transformer architectures
- 🤗 Hugging Face ecosystem
- 🎯 Fine-tuning language models
- 🔎 Advanced RAG
- 🤖 AI Agents
- 🛠️ Tool-using agents
- 🧠 Agent memory
- 🔗 LangGraph
- 🧩 Multi-agent architectures
- 📊 LLM evaluation
- 🔬 AI observability
- 💻 SWE Agents
I like understanding what's happening underneath the frameworks and libraries I use.
Learn
↓
Build
↓
Break
↓
Debug
↓
Understand
↓
Improve
For systems where correctness matters, I prefer separating:
LLM
│
▼
Interpretation
│
▼
Structured Data
│
▼
Deterministic Logic
│
▼
Validation
AI handles understanding.
Code handles correctness.
┌──────────────────────────────────────┐
│ │
│ 🧩 Data Structures & Algorithms │
│ │
│ ⚙️ Backend Engineering │
│ │
│ 🏗️ LLD & System Design │
│ │
│ 🤖 AI Agents │
│ │
│ 🧠 Transformers & LLMs │
│ │
│ 🔎 RAG & Retrieval │
│ │
│ 🤗 Hugging Face & Fine-Tuning │
│ │
│ 🌎 Open Source │
│ │
└──────────────────────────────────────┘
Don't just learn the tool. Understand the system behind the tool.
Algorithms
↓
Data Structures
↓
Software Design
↓
Backend Systems
↓
Databases
↓
Distributed Systems
↓
Machine Learning
↓
Transformers
↓
LLMs
↓
RAG
↓
Agents
↓
Production AI
