Skip to content
View bahramzada's full-sized avatar
๐Ÿฒ
๐Ÿฒ

Highlights

  • Pro

Block or report bahramzada

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please donโ€™t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
bahramzada/README.md

Raul Bahramzada โ€” AI Engineer and Data Scientist

Website LinkedIn Kaggle Email

๐Ÿง‘โ€๐Ÿ’ป About

class RaulBahramzada:
    role      = ["AI Engineer", "Data Scientist"]
    company   = "Kontakt (ABC-Telecom)"
    location  = "Baku, Azerbaijan ๐Ÿ‡ฆ๐Ÿ‡ฟ"
    focus     = ["LLMs & RAG", "Agentic AI", "Computer Vision", "NLP", "Speech-to-Text", "Automation"]
    education = "M.Sc. Business Organization & Management, UNEC"

    def current_work(self):
        return "Shipping production AI systems that turn raw data into decisions."

I build and ship end-to-end intelligent systems: from speech-to-text pipelines and LLM-based evaluation to retrieval-augmented generation, computer vision, and multi-agent orchestration. I care equally about the model and the engineering around it: FastAPI/Node backends, Postgres + pgvector, background jobs, Docker deployments, and clean, testable code. Outside work I build native Apple apps and small products.

  • ๐Ÿ”ญ Building production LLM, RAG, and automation systems as an AI & Automation Engineer
  • ๐Ÿง  Deep interest in agentic AI, retrieval, and applied NLP for the Azerbaijani language
  • ๐ŸŒฑ Exploring evaluation/observability for LLM apps and cost-aware inference
  • ๐Ÿ’ฌ Ask me about RAG, LLM pipelines, computer vision, or automation

๐Ÿ› ๏ธ Tech Stack

Languages

Languages

AI / ML & Data

AI and ML
Transformers YOLOv8 WhisperX RAG pgvector OpenRouter pandas NumPy

Backend, Frontend & Infra

Backend, frontend and infra

๐Ÿš€ Featured Projects

Apps I'm building right now. All are open source and in development.

In development

Agent control room for supervisor-worker AI workflows. A supervisor plans the task, four specialized workers (researcher, strategist, builder, reviewer) execute it in parallel, with live logs, memory, reruns and human-in-the-loop plan review.

FastAPI React Supabase OpenRouter
In development

Native macOS control center for monitoring and managing VPS servers over SSH. Local-first, public-key only, no stored passwords, with a built-in SSH terminal, security diagnostics and a read-only remote file browser.

Swift SwiftUI SSH macOS
In development

A small personal macOS app for playing single-player Windows Steam games on Apple Silicon. A clean native library on top of Wine, with a separate graphics backend per game (D3DMetal, DXMT, DXVK + MoltenVK).

Swift SwiftUI Wine Metal
In development

Azerbaijani micro-habit tracker built around tiny five-minute habits, with mood logging and AI-powered daily tasks. Small steps, big results.

Next.js TypeScript Supabase OpenRouter

๐Ÿ“š More Projects

Project What it does Stack
๐Ÿงฉ MoveMind Self-play Q-learning that exposes per-move Q-values, replays decisions and evaluates saved policies, a fully inspectable RL system React, TypeScript, Express
๐Ÿ”’ Face & Plate Blur Privacy pipeline that detects and anonymizes faces and license plates in images and video with stable tracking YOLOv8, PyTorch, OpenCV
๐Ÿ—ฃ๏ธ Azerbaijani NLP BERT sentiment, NER anonymization and a GPT-2 poetry model in the style of Ramiz RรถvลŸษ™n Transformers, PyTorch
๐Ÿ“ˆ ML Predictors Baku apartment price prediction and a DNN with 96% accuracy on airline passenger satisfaction scikit-learn, TensorFlow
๐Ÿข Selected professional work (production systems, private, details generalized)
Project What it does Stack
๐ŸŽง Multi-Channel QA Evaluation Quality assurance for call-center, chat and social channels: audio โ†’ transcription (WhisperX / Gemini) โ†’ LLM scoring against per-channel rubrics, with reviewer workflows and per-run cost accounting Next.js, Node, PostgreSQL, LLMs
๐Ÿ“š Enterprise RAG Knowledge System Standardized a large internal knowledge base into a citation-aware RAG format (metadata, chunking, source tracking) RAG, Embeddings
๐Ÿงฎ AI SQL Assistant Natural-language (AZ/EN) analytics for non-technical leadership, with an intent router, fuzzy name resolution and safe SQL generation FastAPI, React, SQLAlchemy
๐Ÿ‘ฅ HR Analytics Platform Role-based internal dashboard with an org hierarchy tree, attendance calendars and RBAC-scoped data React, Node, PostgreSQL

๐Ÿง  What I Do

At Kontakt, I design and ship AI systems that run in production, from raw audio and documents to dashboards people use every day. Details are generalized to respect confidentiality.

๐ŸŽง Speech & Conversation Analytics

  • Call-center audio to text in Azerbaijani, Russian and English, with speaker separation and timestamp alignment
  • LLM-based quality scoring against configurable rubrics, with human review and calibration
  • Support for chat and social channels in the same evaluation flow

๐Ÿค– RAG & AI Assistants

  • Multi-agent retrieval-augmented assistants for internal knowledge and for customers
  • Grounded answers that cite their sources and admit when context is missing
  • Natural-language product search over a catalog API, embeddable as a chat widget

๐Ÿ—„๏ธ Natural-Language Analytics

  • Plain-language questions turned into safe, read-only SQL with an intent router

  • Tables and charts for non-technical leadership, with fuzzy name resolution

๐Ÿ—๏ธ Platform Engineering

  • React / Next.js dashboards on FastAPI and Node.js services
  • Role-based access, audit trails, background jobs and cost tracking per run
  • Cloud storage and databases, automated tests and CI for every change

๐Ÿ›ก๏ธ AI Safety & Evaluation

  • Tooling to test chatbots against jailbreaks, prompt injection and prompt leakage
  • Repeatable evaluation workflows for LLM output quality

๐Ÿ”Œ Automation & Integrations

  • Connecting messaging and chat platforms, internal data and reporting tools

  • Bots, scheduled pipelines and workflow automation (n8n)

  • Careful handling of sensitive data: encryption, private storage, least privilege

๐Ÿ“Š Activity

GitHub activity metrics

Yearly commit calendar

๐ŸŽ“ Education, Experience & Certifications

๐Ÿข AI & Automation Engineer Kontakt (ABC-Telecom)
๐ŸŽ“ M.Sc. Business Organization & Management Azerbaijan State University of Economics (UNEC) ยท 2024โ€“2026
๐Ÿงช Data Scientist Program DIV Academy ยท 2025
๐ŸŽ“ B.Sc. Business Management Azerbaijan State University of Economics (UNEC) ยท 2020โ€“2024

Certifications: IBM ยท Python for Data Science, AI & Development ยท DeepLearning.AI ยท AI For Everyone ยท NVIDIA ยท Fundamentals of Machine Learning

"Every dataset has a story. I build the models and the systems to tell it."

bahramzada.az ยท LinkedIn ยท Kaggle ยท Email

Pinned Loading

  1. Face_Plate_Blur Face_Plate_Blur Public

    Jupyter Notebook 1

  2. mask_detection mask_detection Public

    Jupyter Notebook

  3. az-semantic-similarity az-semantic-similarity Public

    Jupyter Notebook

  4. azerbaijani-news-sentiment azerbaijani-news-sentiment Public

    Jupyter Notebook 1

  5. az-ner-blur az-ner-blur Public

    Jupyter Notebook

  6. tiktok-instagram-downloader-bot tiktok-instagram-downloader-bot Public

    Telegram Bot for Tiktok Instagram video download

    JavaScript 1