Azhar Alhasan

AZHAR ALHASAN

Senior Machine Learning Engineer

Malmo, Sweden · Arabic | English | Swedish · contact@azharalhasan.com · +46 70 098 74 90

I ship production GenAI, RAG, and CV systems that handle millions of interactions, improve resolution rates, and lower costs. Recent wins: IKEA's 63-country bot (+35% resolution, -20% support cost), on-prem RAG for 10M+ sensitive docs, and engagement-driving GenAI playlists. I also built mcphub.directory, a directory for Model Context Protocol servers and tools.

Signature outcomes

  • +35% resolution, -20% support cost (IKEA Billie).
  • 10M+ sensitive docs, 20K users (on-prem RAG).
  • +25% engagement, -40% curation effort (GenAI playlists).

What I do

  • Production GenAI: chat/voice, eval loops, latency and cost control.
  • Enterprise RAG: compliance-first retrieval, guardrails, human-in-loop.
  • Edge & CV: on-device analytics, model optimization for constrained hardware.

How I work

  • Discover: goals, constraints, data/infra fit.
  • Prove: eval harness, latency/cost/reliability targets.
  • Ship & learn: serving, autoscaling, guardrails, observability baked in.

Experience

IKEA

Oct 2024–Present · Malmo

Senior Machine Learning Engineer · Retail

  • Led Billie, a 24/7 chat/voice bot across 63+ markets; 8M voice + 2M text + 500K email monthly.
  • Feedback-driven eval and improvement pipelines lifted resolution by 35% and reduced support cost by 20%.
  • Hardened serving for peak loads with caching, autoscaling, and observability (Langfuse/OTel feeding Grafana).

Capgemini

May 2022–Oct 2024 · Malmo

Senior Machine Learning Engineer · Consulting

  • Architected secure, on-prem RAG for Hitachi Energy: fine-tuned open-source LLM over 10M+ sensitive docs for 20K users.
  • Designed retrieval + editing workflows with guardrails, eval harnesses, and human-in-loop review to improve answer quality.
  • Built production GenAI for Spotify: real-time mood-aware playlists; +25% engagement and -40% curation effort via MLOps automation.

Axis Communications

Mar 2019–May 2022 · Lund

Machine Learning Engineer · Video Surveillance

  • Shipped AXIS Object Analytics for detection, tracking, classification across the camera portfolio at no extra cost.
  • Optimized CNNs for ARTPEC SoCs (quantization, memory, throughput) enabling concurrent real-time models; scaled to 76% of cameras.

Core skills

GenAI & RAG

Agentic RAG, embeddings, semantic search, prompt engineering, LangChain, LlamaIndex, MCP integration.

Serving & LLM stack

OpenAI, Anthropic, Google Gemini, xAI Grok; vLLM, Ray Serve, LiteLLM, Ollama; PEFT, LoRA, QLoRA.

MLOps & Infra

AKS, GKE, Terraform, GitHub Actions, Docker Compose; monitoring with Grafana, Langfuse, OpenTelemetry, MLflow; Redis caching.

Data & services

Pinecone, Milvus, Qdrant; PostgreSQL, BigQuery, Azure Cosmos DB, Databricks; Kafka, Google Cloud Pub/Sub.

Backend

FastAPI, Pydantic, Python, JavaScript; API design, eval harnesses, guardrails, routing.

Monitoring & observability

Grafana dashboards, Langfuse traces, OpenTelemetry instrumentation, MLflow tracking for iterative improvement.

Case studies

IKEA Billie — Support Automation

63+ markets; 8M voice/2M text/500K email monthly. Feedback-driven eval, routing, and retraining lifted resolution by 35% and reduced cost by 20%; autoscaling and observability for peak stability.

On-Prem RAG — Hitachi Energy

Fine-tuned open-source LLM; 10M+ sensitive docs; 20K users. Retrieval + editing workflows with guardrails and human-in-loop review; built for compliance and data sovereignty.

GenAI Playlists — Spotify

Real-time mood-aware playlists; +25% engagement and -40% curation effort through scalable MLOps automation and continuous optimization.

AXIS Object Analytics

On-device CV shipped portfolio-wide; optimized CNNs for ARTPEC SoCs (quantization, memory, throughput) enabling concurrent real-time models on constrained hardware.

Education

MSc, Embedded and Intelligent Systems

Halmstad University · 2017–2019

BSc, Electrical and Electronics Engineering

University of Basrah · 2012–2016

Let's talk

Reach out to explore future collaborations—whether you need GenAI, RAG, or ML leadership, or want a quick gut-check on where to take your stack next.

Email meCall or WhatsAppLinkedInGitHub