AI Agents & LLM Systems
Multi-agent orchestration, RAG pipelines, and Mixture-of-Experts architectures. Fine-tuning open-source LLMs and serving them under real concurrency and latency constraints.
I build AI systems that hold up in production.
Software and Machine Learning Engineer with 5+ years architecting scalable backend infrastructure, production-grade AI agents, and high-performance LLM systems.
About
What I work on
I design and ship distributed backend services in Python and Kubernetes, and the AI systems that run on top of them — agent pipelines, RAG architectures, and fine-tuned LLMs serving real users.
My work covers the full delivery path: API and data-model design, database and query optimization, container orchestration on GKE, automated CI/CD, and the production monitoring that catches problems before customers do.
I care about the parts that decide whether a system survives contact with production — latency budgets, concurrency and memory behaviour, graceful degradation, and deployment pipelines fast enough that shipping is not a risk.
Backend Architecture
Distributed services in Django and FastAPI, exposed over GraphQL, REST, gRPC, and SSE. Query optimization, async task pipelines, and caching that keeps response times predictable.
Platform & DevOps
Kubernetes on GKE with ConfigMaps, Secrets, and health probes. CI/CD via GitHub Actions and Cloud Build, plus the observability to run it all with confidence.
Experience
Where I've built things
Software Engineer
Relai AI·Maryland, US (Remote)Current
May 2025 — Present
- Architected scalable AI agent pipelines, integrating a copilot microservice over gRPC and fine-tuning open-source LLMs with Unsloth for domain-specific developer workflows.
- Modernized the API layer by introducing Django Strawberry GraphQL alongside existing REST endpoints, secured end to end with OAuth2 credential integration.
- Built real-time frontend connectivity on Server-Sent Events (SSE) streaming channels, and managed high-concurrency agent memory through shared Redis pools to prevent session bloat.
- Reduced API response latency by 90% through database optimization, resolving N+1 query bottlenecks in the data-access layer.
- Automated the CI/CD lifecycle with GitHub Actions and GCP Cloud Build, cutting deployment turnaround from 20 minutes to 5 via build-stage parallelization and caching.
- Owned end-to-end Kubernetes deployments and backend feature lifecycles, managing ConfigMaps, Secrets, and health probes to keep availability high and downtime minimal.
- Python
- Django
- GraphQL
- gRPC
- SSE
- Redis
- Kubernetes
- GCP
Co-founder, Software & ML Engineer
Contacti AI·Vancouver, BC
Jan 2023 — May 2025
- Led technical execution of conversational AI products, integrating LLMs, STT/TTS, and enterprise-grade RAG pipelines using multi-agent orchestration and Mixture-of-Experts (MoE) architectures.
- Architected a scalable full-stack platform on Django, FastAPI, and Next.js delivering high-performance web applications for complex customer scheduling workflows.
- Engineered omni-channel automation by integrating Twilio for real-time voice and chat, synchronized with internal CRMs and third-party webhook tools including Zapier and Make.com.
- Python
- FastAPI
- Next.js
- LangChain
- RAG
- Twilio
- PostgreSQL
Software & Machine Learning Engineer
Visionish AI·Vancouver, BC
Aug 2022 — Jan 2023
- Deployed production-grade computer vision models for wildlife detection, achieving over 90% F1-score while optimizing for constraint-heavy, real-time edge inference.
- Engineered ETL and automated dataset-scraping pipelines to streamline data ingestion, continuous model training, and vision pipeline scalability.
- Python
- PyTorch
- ONNX
- Edge Inference
- ETL
AI Research Assistant
University of British Columbia·Vancouver, BC
Jan 2021 — Oct 2022
- Authored and open-sourced a safe optimization library, implementing a novel algorithm that accelerated convergence rates from sub-linear to linear.
- Engineered data pipelines for the Raven simulator, applying constraint-relaxation methods to validate industrial optimization algorithms under production-like constraints.
- Python
- Optimization
- Numerical Methods
- Simulation
Skills
Tools I reach for
Languages & Backend
- Python
- TypeScript
- JavaScript
- Rust
- Django
- FastAPI
- GraphQL
- gRPC
- SSE
- Celery
- SQLAlchemy
AI & Machine Learning
- LLMs
- RAG
- AI Agents
- Hugging Face
- LangChain
- LlamaIndex
- LangSmith
- ONNX
- Pinecone
- STT / TTS
Data & DevOps
- PostgreSQL
- MongoDB
- Redis
- Docker
- Kubernetes
- GKE
- GitHub Actions
- CI/CD
- Cloud Build
Cloud Platforms
- AWS
- GCP
- Azure
- DigitalOcean
Background
Education & recognition
Education
Master of Applied Science, Electrical & Computer Engineering
University of British Columbia
Vancouver, BC, Canada · Sep 2020 — Oct 2022
Bachelor of Science, Dual Major in Electrical Engineering & Computer Science
Sharif University of Technology
Tehran, Iran · Sep 2014 — Jul 2019
Awards
Second Place, Beeloud Build AI Hackathon
$1K prize · May 2024
NSERC Research Grant
$12K awarded for the NFRF project · Feb 2021
Languages
- English
- Persian (Native)
Contact
Let's build something
I'm open to senior backend and machine learning roles, plus consulting on AI agent and LLM infrastructure. Drop me a line and I'll get back to you.