Ilkin Guluzada

Ilkin Guluzada 38 records · 3 lit

I embed with your team and ship LLMs to production.

Forward-Deployed Engineer in Baku. LLM agents, RAG and evals, built on twelve years of full-stack engineering.

contact@ilkin-guluzada.comSelected work

About

Ilkin Guluzada, portrait

As a Forward-Deployed Engineer with 12 years of full-stack experience, I work alongside the people who own a workflow to take AI from a real problem to a system running in production. My focus is LLM-based systems: autonomous agents, RAG pipelines, and AI-native products such as Finaps, CreateCV, Scrummy and TryVeo.

I specialize in deploying LLMs into real products, from designing agent architectures and retrieval pipelines to shipping, evaluating and monitoring them in production. Alongside that, I set up CI/CD pipelines, architect scalable front-end systems (Micro Frontends/Monorepos), deliver cross-platform apps for web, mobile, and desktop, and leverage AWS services end to end. A strong UI/UX background keeps my work user-centric, including the interaction patterns AI features need to feel trustworthy.

I measure what I ship: eval sets for answer quality and budgets for latency and cost, so a model or prompt change goes out as a tested release. Proficient in Agile methodologies (Scrum, Kanban), I keep learning fast, which matters in a field where the AI tooling changes monthly.

12
years building software
100K+
developers taught on Udemy
AWS Certified Cloud Practitioner
certification
BSc Computer Science, MBA
ADA University [TBD] degrees

What I do draft copy

  • AI & LLM engineering

    Agents with tool calling, retrieval pipelines (vector search, embeddings) and structured outputs, designed around the workflow they serve rather than a demo.

    OpenAI and Anthropic APIs, RAG, MCP integrations, prompt engineering, multi-model generation

  • LLM deployment & evals

    Getting a model into production and keeping it there: CI/CD that runs evals alongside the tests, containerized AI services, and monitoring for latency, cost and answer quality.

    GitHub Actions, Docker, Amazon Bedrock, eval sets, human-in-the-loop approvals

  • Web & mobile

    Next.js and React front ends on Node services, and cross-platform apps for web, mobile and desktop: the product around the model.

    Next.js 14, React, TypeScript, React Native, Node.js (Express, Nest), PWAs, Electron

  • APIs & data

    tRPC, REST and GraphQL over relational, document and vector stores, plus the integrations that give an AI feature real data to act on.

    tRPC, REST, GraphQL, PostgreSQL with pgvector, MySQL, MongoDB, Prisma

  • Cloud & DevOps

    Serverless and containerized services on AWS, with CI/CD pipelines that automate testing and deployment.

    AWS (Lambda, DynamoDB, S3, Cognito, Bedrock), Docker, GitHub Actions, GitLab CI/CD, CircleCI

  • Product & UI

    From Figma to production, including the interaction patterns AI features need to feel trustworthy: drafts you approve, sources you can check.

    Figma, design systems, Tailwind, Material UI, UI/UX research

Selected work

Four AI products, live in production.

  1. Finaps (opens in a new tab)

    finaps.co

    Personal finance app: every account in one dashboard, budgets kept in envelopes, receipts parsed from a photo.

    Finaps Agent drafts the action, such as cancelling an unused subscription or splitting a bill, and waits for your approval; every step is logged and reversible.

    Finaps homepage

  2. CreateCV (opens in a new tab)

    createcv.io

    Resume builder with ATS-optimized templates, an ATS checker that scores each CV, and job search that applies for you, assisted or automatic.

    Smart AI Apply tailors the resume to each job description and brings the matching skills to the top.

    CreateCV homepage

  3. Scrummy (opens in a new tab)

    scrummy.pro

    One platform for agile ceremonies: standups, retrospectives and planning poker, connected to the sprint data in Linear and Jira.

    AI Scrum Master chats about the sprint data, flags recurring blockers and turns retrospectives into action items, as a coach, a challenger or an executive.

    Scrummy homepage

  4. TryVeo (opens in a new tab)

    tryveo.ai

    Creative studio for video, images, music and voice: type an idea, pick a named effect, get the result. In Azerbaijani, English and Russian.

    135+ AI models sit behind one prompt, from Veo 3.1 for video to Suno for music.

    TryVeo homepage

Ask the index

Ask in your own words. The index returns the passages from this page that match, quoted as written, and names where each came from. There is no model behind it yet. Your question stays in this browser.

v1 retrieval passes 20 of 20 questions on the public eval set, measured at build.

0 / 200

Experience

newest first

  1. 2023 to now

    Senior Software Engineer at Grapevine Tech

    • Architected React and Next.js applications in TypeScript and GraphQL, including LLM-powered features and AI agent workflows integrated into production
  2. 2022 to 2024

    Frontend Coach at Jagaad Academy

    • Developed and delivered a comprehensive frontend curriculum covering HTML, CSS, and JavaScript, improving student project completion rates by 40% over 6 months.
  3. 2021 to 2023

    Lead of Front-end Engineering at Pasha Bank

    • Architecting React, Next.js applications with GraphQL and TypeScript
  4. 2019 to 2021

    Lead Full-Stack Engineer at Travolic LLC

    • Reviewing codes and confirming pull requests in Github
  5. 2017 to 2019

    Senior Software Engineer at Azerconnect LLC

    • Determining the structure and design of web pages
  6. 2015 to 2017

    Lead Software Engineer at CStreet Media Inc.

    • Managing team of 4 developers
  7. 2014 to 2015

    Software Engineer at CRBN Tech

    • Researching, designing, implementing and managing software programs

Stack

Pick a facet to re-rank the index and the roles above by where it was used.

AI & LLM engineering

OpenAI and Anthropic APIs, AI agents and multi-step workflows, Tool calling, RAG (vector search, embeddings), Prompt engineering, Structured outputs, MCP integrations, Multi-model generation (video, image, music, voice)

LLM deployment & evals

Eval sets and regression checks, Latency, cost and response-quality monitoring, CI/CD that runs evals (GitHub Actions), Containerized AI pipelines, Amazon Bedrock, Compute and API budgets for AI workloads, Human-in-the-loop approvals

Web & mobile

JavaScript (ES6+), TypeScript, React (Next.js 14, Redux, React Query, Zustand), React Native (iOS/Android), Node.js (Express.js, Nest.js), Progressive Web Apps (PWAs), Python for AI tooling and data pipelines. Tailwind, Material UI, Bootstrap. Webpack, Vite, Turbopack. Shopify Hydrogen

APIs & data

tRPC, REST, GraphQL (TypeGraphQL). PostgreSQL (with pgvector), MySQL, MongoDB, Vector databases, Prisma, TypeORM, Sequelize, Mongoose

Cloud & DevOps

Docker, AWS (Lambda, DynamoDB, S3, Cognito, Bedrock), Serverless Architecture, CI/CD (GitHub Actions, GitLab CI/CD, CircleCI), DigitalOcean, On-Premise

Product & UI

Figma, Adobe XD, Photoshop, Illustrator, Visual research, Wireframing, Sketching, Design systems, UI/UX-led product thinking

Practices & tooling

Agile (Scrum, Kanban), Micro Frontends / Monorepos, Algorithms & data structures. GitHub, GitLab, Bitbucket, Jira, Trello. Electron (desktop apps), VS Code Extensions, Lerna, NPM, TSOA, JAMstack, Headless CMS (Contentful)

Teaching

100K+ developers taught on Udemy (opens in a new tab), four courses.

  • MERN Stack - Complete BlogWeb development
  • Mastering UI/UX using FigmaUI/UX
  • Mastering HTMLFront-end development
  • Mastering DenoBack-end development

YouTube channel Ilkin Dev (opens in a new tab), for example "Mastering Figma: What is Figma?"

Frontend Coach at Jagaad Academy, 2022 to 2024: mentored 30 junior developers, 80% placed within 3 months.

Writing

Archive, 2019 to 2020. All posts