Reza
Dorkhah_
AI Engineer
5+ years shipping production-grade LLM, RAG and agentic systems for industrial monitoring, banking, insurance and healthcare — from vector pipelines to deployed agents.
5+
Years production AI
4
Industries
6
Certifications
NLP
+ CV Focus
01 / About
Currently building
AI-assisted industrial monitoring on AWS Bedrock — retrieval over signals, weather, assets and operational data.
Stack in production
LangChain · AWS Bedrock · FAISS · FastAPI · PyTorch · Docker · ClearML
I build AI that survives contact with production.
As an ML Engineer at Hifi Engineering, I take LLM and RAG systems from prototype to monitored, deployable infrastructure — evaluation, latency, and the unglamorous details that decide whether a system earns trust. My background spans data science at ShoreWise Consulting and Statistics Canada, plus regulated sectors where reliability is non-negotiable.
LLM orchestration, tool calling & structured outputs
RAG pipelines: chunking, embeddings & vector stores
Agentic workflows with human-in-the-loop review
Evaluation harnesses, observability & MLOps
02 / Experience
Machine Learning Engineer — Hifi Engineering
Nov 2025 — PresentCalgary, AB
Building an AI-assisted industrial monitoring application on Amazon Bedrock that orchestrates retrieval and analysis tools across signals, weather, assets, documents and operational data. Trained a custom embedding model over historical spectrograms with FAISS similarity retrieval, designed human-in-the-loop feedback workflows, and architected a hybrid on-prem/AWS MLOps platform with ClearML, SageMaker, Docker and shadow-testing.
Data Scientist — ShoreWise Consulting
Mar 2021 — Oct 2025Remote
Designed and delivered production NLP, Generative AI, RAG and agentic solutions. Built an on-premises banking RAG system over sensitive documents with open-weight LLMs, document-processing pipelines combining OCR and vision-language models, and multi-step agentic workflows with structured-output validation, retries, fallbacks and human escalation.
Data Scientist Co-op — Statistics Canada
May 2022 — Dec 2022Ottawa, ON (Remote)
Built data pipelines and analytical workflows in Python and R for large-scale datasets, evaluated machine-learning models through iterative experimentation, and delivered insights to technical and non-technical stakeholders.
Capabilities
LLM & Generative AI
LangChain, LangGraph, RAG, prompt engineering, tool calling, structured outputs, open-weight LLMs, QLoRA, Hugging Face
Retrieval & Knowledge
Embeddings, FAISS, vector databases, semantic search, metadata filtering, context construction, document retrieval
Evaluation & Reliability
Ground-truth evaluation, groundedness and failure analysis, FP/FN analysis, human feedback, shadow testing, latency evaluation
Backend & APIs
Python, FastAPI, Flask, Pydantic, REST APIs, AWS Bedrock, Lambda, Redis, MongoDB, PostgreSQL
ML & Data
PyTorch, Scikit-learn, Pandas, NumPy, PySpark, SQL, transformers, data pipelines
MLOps & Infrastructure
ClearML, SageMaker, Docker, Kubernetes, MinIO, S3, EC2, experiment tracking, monitoring, retraining
03 / Selected Work
Live insurance comparison platform for Oman. Conversational intake, a vector store of real insurer quotes, and semantic matching that ranks options against customer requirements — with deterministic eligibility rules, structured validation, and human handoff for regulated questions.
Live practice-management platform for health and wellness practitioners. AI-powered scheduling, session plans, clinical summaries and reports — LLM workflows that turn shorthand notes into structured, review-ready documentation.
Live AI training platform for coaches, athletes and the self-coached. Personalized coaching, live cues and shared progress built on wearable data — adaptive plans and multi-language delivery, end to end.
Education
MSc Professional Computer Science
Simon Fraser University
BSc Electrical Engineering
Sharif University of Technology
Certifications
04 / Contact