# Diwash Bhandari

> Senior Software Engineer | Python, FastAPI, AI/ML & System Architecture

Kathmandu, Nepal · 5+ Years · Open to Opportunities

## Summary

I build backend systems, AI applications and SaaS platforms that keep running as they grow.

## Highlights

- Built scalable REST APIs, GraphQL APIs, and real-time backend services for cloud telephony and SaaS platforms
- Integrated CRM, customer support, marketing automation, analytics, and communication platforms using REST APIs, GraphQL, webhooks, ETL pipelines, and event-driven architectures
- Developed AI-powered applications using RAG pipelines, LangChain, LangGraph, vector databases, and multiple LLM providers
- Built AI products end to end: IntelliDocs AI answers questions from uploaded documents, and WACRM (in progress) adds an AI reply agent to a WhatsApp CRM
- Designed secure authentication systems, real-time data synchronization pipelines, and scalable microservices using FastAPI, PostgreSQL, Redis, and EdgeDB (Gel)
- Improved engineering quality through code reviews, mentoring, SOLID principles, CI/CD practices, and cross-functional collaboration

## Core Competencies

- Backend Architecture & API Design
- Cloud Telephony & CRM Integrations
- Microservices & Distributed Systems
- AI/LLM & RAG Applications
- Event-Driven Architecture & Real-Time Systems
- Performance Optimization
- DevOps & CI/CD

## Services

### Backend API Development

From $2,500 · 2-4 weeks

High-performance REST and GraphQL APIs built with FastAPI, handling 100K+ requests/day with sub-200ms response times. Includes authentication, authorization, database optimization, and comprehensive API documentation.

- RESTful API design & development
- GraphQL APIs with subscriptions
- Database architecture & optimization
- Authentication & authorization systems
- API documentation & testing

**Technologies:** FastAPI, PostgreSQL, Redis, GraphQL, Docker

### AI/ML Integration

From $3,500 · 3-6 weeks

Production-ready AI applications — RAG platforms, LLM-powered chatbots, and document Q&A systems with 92% accuracy. Built for enterprise teams needing fast, reliable AI-powered responses.

- RAG (Retrieval-Augmented Generation) systems
- LLM-powered chatbots & assistants
- Document processing & Q&A
- Vector database integration
- Multi-provider AI orchestration

**Technologies:** LangChain, OpenAI, ChromaDB, Qdrant, FastAPI

### System Architecture & Consulting

From $1,500 · 1-3 weeks

Microservices design, monolith-to-microservices migration, and performance optimization. Reduced deployment time by 40% and scaled systems from 10K to 100K+ users with containerized infrastructure.

- Microservices architecture design
- Monolith decomposition
- Performance optimization
- System scalability planning
- Code review & mentoring

**Technologies:** Docker, AWS, gRPC, RabbitMQ, PostgreSQL

### Cloud Telephony Solutions

From $3,000 · 3-5 weeks

IVR workflows, call routing, SMS automation, and CRM integration across multiple countries with real-time monitoring dashboards.

- IVR system design & implementation
- Call routing & recording
- SMS & notification automation
- CRM integration
- Real-time call monitoring dashboards

**Technologies:** Twilio, DIDWW, Telynx, WebSockets, FastAPI

## Experience

### Senior Software Engineer — Freelance

May 2026 - Present · Remote · Freelance / Consulting

Design, build and run backend systems for clients: warehouse operations, a large loyalty program, software licensing and AI customer support.

- Built a warehouse backend shared by several clients, with live slot locking and dashboards.
- Built a member platform for about 30M records, with roles, caching and store suggestions from a machine-learning model.
- Built a license server and a matching client library that check signed (Ed25519) licenses and keep working offline for a set grace period.
- Shipped NepalConnect: an AI support agent with 24 tools for an internet provider, plus customer setup through FreeRADIUS, document search and payments through eSewa and Khalti.

**Technologies:** Python, FastAPI, PostgreSQL, Redis, Docker, AI/ML, WebSockets

### Mid-Level Software Engineer — Codavatar Tech Pvt. Ltd.

June 2023 - May 2026 · Kalopul, Kathmandu · Full-time

Built the backend for cloud phone and CRM products: secure APIs, real-time updates and integrations that let teams automate customer calls and messages.

- Built real-time GraphQL subscriptions over WebSockets, so apps no longer had to keep polling for changes.
- Led the move to Casdoor for login and moved existing user accounts across for business-critical services.
- Designed a plug-in style CRM connector that several product teams reused.

**Technologies:** Python, FastAPI, EdgeDB (Gel), PostgreSQL, GraphQL, Redis, Twilio, DIDWW, Telnyx, Casdoor

### Software Engineer — Chuchuro Firm

May 2022 - July 2023 · Sinamangal, Kathmandu · Full-time

Made the system faster and more reliable by moving slow work to background jobs, adding caching and running every environment in Docker.

- Moved slow tasks to background jobs with Celery and RabbitMQ, so API responses came back faster.
- Added Meilisearch to make full-text search faster and more accurate.
- Set up the same Docker environment for development and staging.

**Technologies:** Python, Tornado, Celery, RabbitMQ, Redis, PostgreSQL, Peewee ORM, Meilisearch, Docker

### Associate Software Engineer — Young Minds Creation (P) Ltd.

December 2020 - April 2022 · Young Minds Tower, Kathmandu · Full-time

Built and shipped web applications for business clients that stayed up reliably and kept clients happy.

- Named Eminent Developer of the Year for my contributions to the company's software and new ideas.
- Built and shipped web applications for business clients that stayed up reliably and kept clients happy.

**Technologies:** PHP, Laravel, MySQL, Docker, REST APIs, Middleware

## Projects

### Warehouse Management System (WMS)

Warehouse backend for many clients — receiving, stock, picking and live slot updates · Backend Systems | Warehouse Operations

A backend that runs warehouse work for several clients from one system: receiving goods, tracking stock, picking orders, managing storage slots and loading. Each client's data lives in its own PostgreSQL schema. When a worker reserves a slot, Redis and WebSockets push the change to every screen right away, so two people can't take the same slot. It exchanges files with outside systems over SFTP and CSV, and has dashboards for traffic, pick logs and audit history.

**Technologies:** Python, FastAPI, PostgreSQL, SQLAlchemy, Redis, WebSockets, SFTP, CSV, Sentry

### Member Management System (MMS)

Member platform for a loyalty program — about 30M records, roles, caching and store suggestions · Backend Systems | Enterprise SaaS

A backend for a loyalty and club program that handles member sign-up, status changes, card transactions and club summaries across very large tables. Keyset pagination and Redis caching keep big lists fast. Every member change is logged, access is controlled by roles, and documents are stored in S3. Sync jobs connect it to outside systems, and a scikit-learn model suggests good store locations.

**Technologies:** Python, FastAPI, Async SQLAlchemy, PostgreSQL, Redis, S3, scikit-learn, Claude API, Twilio, SFTP, CSV, Sentry

### Guard License Server

License server — signed licenses that work offline, locked to one machine or shared · Backend Systems | Security | Licensing

A server that creates and checks software licenses. Each license is signed with Ed25519, so an app can verify it without calling the server. Licenses can be locked to one machine or shared between machines (floating). Data can be stored in MessagePack files, SQLite or PostgreSQL and is encrypted with AES-256-GCM, and the admin API has rate limits and structured logs.

**Technologies:** Python, FastAPI, PostgreSQL, SQLite, MessagePack, AES-256-GCM, Ed25519, Docker, Loguru

### Guard Trust — License Enforcement SDK

Python library that checks an app's license on the user's machine, even offline · Backend Systems | Security | Licensing

A library that apps include to check their own license. It verifies the Ed25519 signature locally and matches the license to the machine using its CPU, network card, disk and hostname. If the license server is down, the app keeps working for a set grace period. It checks the license at startup and in the background and notices when a license is revoked. A FastAPI middleware re-checks it on requests, and Nuitka compiles the code so it is hard to read or change.

**Technologies:** Python, FastAPI, Ed25519, AES-256-GCM, PostgreSQL, Nuitka, Docker, GitLab CI/CD

### NepalConnect

Support and automation platform for an internet provider — AI agent, network control and payments · Backend Systems | Telecom & AI Support

A platform that answers customer questions and handles routine tasks for a broadband provider. A LangGraph AI agent can use 24 tools, and LiteLLM switches between three model providers (OpenAI-compatible, Groq and Gemini) when one fails. Presidio removes personal data before it reaches a model, and answers come from a search that mixes vector and keyword results. It connects to FreeRADIUS and GenieACS to set up new customers, change plan speeds instantly, slow down users over their data limit and check routers remotely. Payments through eSewa, Khalti, FonePay and ConnectIPS are processed exactly once and reconciled every 5 minutes.

**Technologies:** Python, FastAPI, LangGraph, LiteLLM, PostgreSQL, Upstash (Vector/QStash), Redis, FreeRADIUS, Docker, CI/CD

### IntelliDocs AI

Upload documents and ask questions about them — answers come back with their sources · AI Systems | Document Q&A

A backend where teams upload documents (PDF, Word, images or text) and ask questions about them. A background pipeline built with LangGraph, Celery and RabbitMQ reads each file with the right tool (PyMuPDF, Unstructured, python-docx or Tesseract OCR) and splits it into searchable pieces. Each customer's documents are stored in a separate Chroma collection, so one customer never sees another's results. Files are checked for hidden text, macros and leaked passwords before they are saved, and Presidio masks personal data. Answers stream back with their sources, the app switches between Gemini, Cohere and Groq if one fails, and login uses Keycloak.

**Technologies:** Python, FastAPI, LangGraph, LiteLLM, ChromaDB, LangChain, PostgreSQL, Celery, RabbitMQ, Keycloak, Presidio, Docker

[Live](https://intellidocs.diwash-bhandari.com.np/)

### WACRM — WhatsApp CRM

WhatsApp CRM with an AI reply agent — shared inbox, sales pipeline and broadcasts · AI Systems | WhatsApp CRM · In progress

A self-hosted CRM for teams that sell and support customers on WhatsApp, built on an open-source WhatsApp CRM template with Next.js and Supabase. My work adds the AI reply agent. It searches the team's knowledge base four ways at once (exact numbers, meaning with pgvector, keywords with Postgres full-text search and typos with pg_trgm) and merges the results. Order, invoice and ticket numbers are always looked up exactly, so the agent never answers about the wrong order, and customers can ask for the status of their own deals. A short questionnaire writes the agent's instructions with fixed privacy rules, and a leak guard stops any reply that contains API keys or internal instructions and hands the chat to a person. I also redesigned the login and dashboard screens.

**Technologies:** Next.js, TypeScript, Supabase, PostgreSQL, pgvector, React, Tailwind CSS, Vitest

## Skills

- **Programming Languages:** Python, Go, JavaScript
- **Backend Frameworks:** FastAPI, Starlette, Laravel, Tornado
- **Ai Ml Technologies:** LangChain, OpenAI GPT, ChromaDB, FAISS, RAG Systems
- **Databases:** PostgreSQL, Redis, MongoDB, MySQL
- **Apis Architectures:** RESTful APIs, GraphQL, gRPC, Microservices
- **Devops Cloud:** Docker, AWS, Git, CI/CD
- **Frontend:** React, HTML/CSS
- **Other Technologies:** WebSockets, AsyncIO, TDD

## Education

- **Bachelor in Information Management** — [Nepal Commerce Campus (NCC)](https://ncc.tu.edu.np/) (Information Technology & Business Management · 2017 - 2021)
- **Higher Secondary Education (Management - Computer Science)** — [Ambition College](https://ambition.edu.np/) (Computer Science & Business Studies · 2014 - 2016)

## Writing

- [Context Engineering: The Next Evolution Beyond Prompt Engineering](https://medium.com/@diwasb54/context-engineering-the-next-evolution-beyond-prompt-engineering-94f0ea3c0f75) — Medium · January 2026. In-depth article exploring the shift from prompt engineering to context engineering for building more effective AI systems
- [MVP vs SLC: Which Development Strategy Wins for Web Projects?](https://medium.com/@diwasb54/mvp-vs-slc-which-development-strategy-wins-for-web-projects-2f405ef46921) — Medium · December 2025. Comparative analysis of Minimum Viable Product vs Simple, Lovable, Complete development strategies
- [Pydantic v2: 24 Features That Will Transform Your Python Code](https://medium.com/@diwasb54/pydantic-v2-24-features-that-will-transform-your-python-code) — Medium · December 2025. Comprehensive guide to new features in Pydantic v2 with practical examples and use cases

## Certifications

- [Python For Beginners Course In-Depth](https://www.udemy.com/certificate/UC-8b76db5f-747b-47b2-9224-7f2f2cc79a0d/) — Udemy · October 2022
- [The Complete ReactJs Course - Basics to Advanced [2021]](https://www.udemy.com/certificate/UC-9c90c56a-2f49-41db-9b8b-2a88aef3ffc2/) — Udemy · August 2021

## Awards

- **Eminent Developer of the Year** — Young Minds Creation (P) Ltd · February 2022. Awarded on the 15th Annual Day for exceptional contributions to software development, innovation, and leadership in tech projects

## Contact

- **Email:** dev@diwash-bhandari.com.np
- **Phone:** +977-9861037328
- **Location:** Kathmandu, Nepal
- **Availability:** Open to Opportunities
- **Contact form API:** POST https://www.diwash-bhandari.com.np/api/contact
- **LinkedIn:** https://www.linkedin.com/in/diwash-bhandari-3814b2157/
- **GitHub:** https://github.com/Diwas2055/
- **GitLab:** https://gitlab.com/Diwas2055

## Machine-Readable Resources

- OpenAPI 3.1 specification: https://www.diwash-bhandari.com.np/openapi.json
- API documentation: https://www.diwash-bhandari.com.np/api/docs
- MCP manifest: https://www.diwash-bhandari.com.np/.well-known/mcp/manifest.json
- MCP endpoint (Streamable HTTP, JSON-RPC 2.0): https://www.diwash-bhandari.com.np/mcp
- llms.txt: https://www.diwash-bhandari.com.np/llms.txt
- Sitemap: https://www.diwash-bhandari.com.np/sitemap.xml

---

- Canonical URL: https://www.diwash-bhandari.com.np/
- Source: Diwash Bhandari (www.diwash-bhandari.com.np)
- This markdown is generated from the same data as the HTML page at this URL.