Skip to content

Kushal Gajurel Lead engineer Kathmandu, Nepal

End-to-end products for the agentic era.

I build the whole product: Python APIs and data pipelines, React and Next.js front ends, and the LLM and agent workflows in between. I chase the projects that make someone's day easier.

  • Building software since 2017
  • GMT+5:45, overlapping EU and US mornings
  • Open to 4 to 8 week scoped engagements
Currently building at

Selected work

A few things I've built.

More on GitHub

IZY Prediction at Resimator.fi

Led backend engineering for IZY Prediction at Resimator.fi, delivering scalable APIs and data pipelines for production prediction workflows.

Technologies: Python · FastAPI · Backend · Data Pipelines · Prediction · Lead Engineer

CogniStream.ai

Led engineering for AI-powered voice research workflows, including LLM-assisted analysis pipelines and backend product integrations.

Technologies: AI · LLM · Agentic Systems · Backend · Voice Research · Product Engineering

Vritta

Self-hosted log normalisation for SOC and MSSP teams. It identifies each log stream, maps it to OCSF, and drafts a parser for new sources that an analyst approves before it goes live.

Technologies: Next.js · OCSF · Log parsing · ClickHouse · Security · Self-hosted

Artisai

Website for Artisai, an independent software agency in Ireland. I helped build the site that presents its result-as-a-service model and selected client work.

Technologies: Next.js · Web design · Vercel

Ryuk AI

Ask a database in plain English. Ryuk writes and runs read-only SQL, then returns the result with the query, source rows and a chart, ready to save as a metric or share in a report.

Technologies: Next.js · LLM · Text-to-SQL · PostgreSQL · Turso

Wave

GPR Software's WAVE turns raw ground-penetrating radar data into accurate 3D models, keeping full resolution and geolocation while processing large surveys quickly.

Technologies: GPR · 3D modelling · Data processing · Geophysics

Speaker Recognition

Python project using MFCC for voice-based identification and recognition with Artificial Neural Network.

Technologies: Python · PyQt5 · numpy · scipy · sklearn · MFCC · ANN · Principal Component Analysis

Violence Analysis

Applied NLP on Nepali news data for violence analysis, extracting insights through tokenization.

Technologies: Python · NLP · Tokenization

A little about me

Different stops.
Same instinct.

View my résumé

I write backend systems for a living, though it took a while to land here. I've been a security analyst, a data analyst, and now I lead backend engineering, and every stop along the way still shows up in how I build.

Right now I'm the lead backend engineer at Resimator.fi, building the Python APIs and data pipelines behind IZY Prediction, and I lead engineering at CogniStream.ai, where we're using LLMs to make voice research faster. Day to day that's Python and FastAPI, React and Next.js when the job needs a frontend, and lately a lot of LangChain, LangGraph, and RAG pipelines. The agentic side of the stack is where most of the interesting problems are right now.

Before backend was the job, security was. I spent years in a SOC doing incident response and log analysis, and building things that had to hold up under CCPA and GDPR, not just work in a demo. That part of my background doesn't show up on a typical developer's resume, but it's probably the most useful thing I carried over: I still build assuming something is trying to break it.

Ask what that means day to day and the honest answer is: I write code for two very different readers. One skims it once and moves on. The other parses every line, retries the parts it doesn't understand, and never says thank you.

Beyond code

Music
Rapping and producing, when there's time for it.
Listen on Spotify
Media
A handful of creative media projects on the side.
Creative media
Digital art
Generative AI and VFX experiments.
Animation
Motion graphics, slowly getting better at it.

Different mediums, same instinct: build the thing properly, then see if it holds up.

Experience

The path so far.

Lead Backend Engineer (Senior Backend)2023 to present Remote

At Resimator.fi, I led backend architecture and delivery for IZY Prediction, building and scaling Python APIs, data pipelines, and production services for prediction workflows.

Lead Engineer2024 to present Remote

At CogniStream.ai, I lead engineering for AI voice-research workflows, including backend systems, LLM-assisted automation, and product integrations for faster qualitative insights.

Full-Stack Developer2023 to present Kathmandu, Nepal

I'm currently working as a Full-Stack Developer on my own projects. My stack includes React, Next.js, TypeScript, Tailwind, and MongoDB. I'm open to full-time opportunities.

Senior Python Developer01/11/2023 to present Kathmandu, Nepal

Developed Flask-based backend for a hospitality product, collaborated on FastAPI integration with Stripe Connect for secure payments, ensuring compliance and streamlined processing.

Data Analyst01/05/2022 to 01/11/2023 Kathmandu, Nepal

Worked in database management, data modeling, ELT processes, and CI/CD pipelines. Utilized AWS and Azure for application deployment and data workflows. Conducted data analysis and created interactive visualizations.

Security Analyst01/05/2020 to 01/05/2022 Kathmandu, Nepal

Developed a Python log analysis framework for log analysis and performance optimization. Demonstrated expertise in compliance regulations (CCPA, GDPR) and collaborated with cross-functional teams to enhance security controls.

Associate Security Analyst01/12/2017 to 01/05/2020 Kathmandu, Nepal

Worked in a Security Operations Center (SOC) environment, monitoring security systems and tools to detect threats and vulnerabilities. Participated in incident response activities and analyzed security incidents.

Skills & capabilities

What I bring
to the table.

Audit & Strategy

Before the first line of code, the API has to be readable by something that doesn't read like you do. I look at what's there and tell you what an agent will trip over.

API audits for agent-readiness

Idempotency, error semantics, schema clarity, auth surfaces, retrievability, audited end-to-end with a remediation plan an engineer can act on tomorrow.

IZY Prediction backend at Resimator.fi

MCP server hardening

Security and performance review of existing MCP servers: rate limits, sandboxing, prompt-injection defense, tool-surface minimization.

Build & Migrate

Most APIs were built for a human reading a Postman collection. Migrating them for agents is a different problem with the same shape.

MCP migration

Wrap or refactor existing REST/gRPC/SOAP APIs into MCP servers: schema design, transport selection, auth bridging, deploy.

MCPifying the World: Field Report

Greenfield MCP server design

New MCP servers from scratch, with tool surfaces shaped for the agents that will actually call them.

MCPifying the World: Workflow-Level Tool Design

Agent infrastructure

Tool-calling backends, retrieval pipelines, agent state, evaluation harnesses, observability. The plumbing under the agent.

LLM voice research workflows at CogniStream.ai

Quality & Adoption

Once it's shipped, you need to know how it behaves, for both readers. And you need a team that can ship with the new tools without making everything worse.

Evals & observability for agents

Evaluation suites and traces for agent-consuming systems, the closest thing to a test runner you'll get in this era.

Auditing AI Decisions: Industry Framework

AI-assisted dev consulting

Help teams adopt Claude Code / Cursor workflows, author internal agent skills, set up agentic CI without turning the codebase into a science fair.

Tools & technologies

Frontend
HTML, CSS, React, Next.js, Tailwind, Framer Motion, Ant Design
Programming Language
Python, Rust, JavaScript, TypeScript, C++, R Language
Cloud
Azure AD, AWS CodeDeploy, AWS CodePipeline, Azure Pipelines
Backend
Flask, FastAPI
AI & Agentic
LangChain, LangGraph, AI Agentic Building, RAG Pipelines, Prompt Engineering
AI Tools
Codex (OpenAI), Claude (Anthropic), Magic, Path, Stitch (Google)
Database
MongoDB, PostgreSQL, MSSQL, Neo4j, SQLite
Tools
Git, Visual Studio Code, SonarQube, GitHub Actions, VMware, Active Directory, Postman, Wireshark, Docker, Jira

Have something in mind?

Let's build something.

Open to one or two engagements at a time. If any of this sounds like the work you need, the fastest way in is email. I read everything; I reply to most things within two days.

Say hello

kushalgajurel@icloud.com

Or gajurelkushal1994@gmail.com

Working together

Taking on first dedicated agent-era engagements. Happy to discuss audits at no charge.

Best fit: 4 to 8 week scoped work, async-friendly, GMT+5:45 base with overlap for EU/EST mornings.

I'm open to full-time opportunities.