Work Experience

Lead backend and AI systems development for a US medical toxicology company, building production software and AI-enabled services for clinicians. Own ToxLumen — the company’s AI-enabled clinical toxicology platform — end to end, from backend architecture and clinical AI workflows to deployment and reliability in production.

• Own ToxLumen end to end: an AI-enabled clinical toxicology platform that takes a case from multimodal intake — typed, recorded, or a live transcribed call — through transcription, structured extraction, and LLM-assisted analysis to a finalized clinical note.
• Design and maintain the backend architecture: Django/DRF services, asynchronous pipelines on Huey and Celery, Redis, and PostgreSQL, under a strict services/selectors layering with transactional writes and an audit trail behind every state change.
• Build clinical AI workflows on LLM and speech APIs (OpenAI, Gemini, Groq, Deepgram): transcription, structured clinical fact extraction, analysis, treatment guidance, and safety alerts — AI assists, the clinician decides.
• Consolidated four production apps — MedSpeech, Simulation, ToxPubMed, and a standalone app per calculator — into one product with one identity, one billing surface, and one design system.
• Run production deployment and reliability: reproducible, containerized releases with Docker, Nginx, and GitLab CI/CD, with a production-readiness audit before the first real users.
• Designed the company’s multi-service estate — independent Auth, CoreLogic, Management, CMS, and Support services — with five-tier RBAC across dedicated dashboards.
• Lead the engineering team: technical roadmap, sprint planning, incident response, and cross-functional coordination with clinicians.

Own the API backend and the production infrastructure of a Persian e-commerce platform for 3D printing, serving a web storefront, an admin dashboard, and a Bale/Telegram bot from a single Django service.

• Built and own the Django/DRF backend — fifteen domains, 53 models and ~191 endpoints — under a layered pattern where services own writes and selectors own reads.
• Enforced a 93% coverage gate on a ~4,700-test suite, wired into the deploy workflow so a coverage drop or a missing migration stops the release instead of shipping.
• Built the observability stack from scratch: Prometheus, Alertmanager and nine exporters behind a custom dashboard — two of them written in-house after cAdvisor image registry proved unreachable from Iran.
• Led the response to a ransomware incident on internet-exposed datastores: rebound PostgreSQL and Redis to loopback, moved credentials to environment-driven config, and audited every published port across the compose files.
• Built an off-site backup pipeline with a documented restore runbook — PostgreSQL dumped to object storage with a SHA-256 manifest per run, retention pruning, and per-run reporting to the team.
• Set up SSH-based GitHub Actions deployment for five independent repositories, each gated on its own CI.

Co-founded an Iranian fitness platform and own its technology end to end — the backend, the gym-owner panel, the superuser dashboard, and the ongoing migration of the production Django system to NestJS.

• Designed and built the Django/DRF backend — 38 models and ~152 endpoints across public, owner and admin scopes — with ownership enforced by scoping the query, so a foreign record returns 404 rather than 403 and never discloses that it exists.
• Built the training domain on immutable snapshots: a workout session freezes exercise names, reps and targets at start, so session history survives later edits or deletion of the plan behind it.
• Built the gym-owner panel and the superuser admin SPA in React 19 + Vite with TanStack Router and Query, both covered by Vitest suites.
• Leading the migration to a NestJS + Prisma backend as a parity contract rather than a rewrite — phase gates, an endpoint-level parity matrix backed by evidence, and a numbered register of every approved deviation.
• Shape product scope and roadmap alongside the engineering work as co-founder, deciding which surfaces get built and in what order.

Designed and delivered backend systems and mobile integrations across multiple client projects. Focused on maintainable, secure, and production-ready architectures tailored to each domain.

• Built high-performance REST APIs and async task management systems using Django/DRF, Celery, and Redis.
• Delivered consistent containerized deployments across environments using Docker and Docker Compose.
• Collaborated with Flutter front-end teams to integrate backend services into mobile clients seamlessly.
• Delivered projects across e-commerce, rental marketplaces, and university administrative systems.

Completed the backend for QOK, a real-world online mobile game, as part of a professional internship. Gained first production experience building scalable backend systems under senior mentorship.

• Delivered full backend for QOK, an online mobile game — first production deployment milestone.
• Worked across Django, DRF, Celery, PostgreSQL, Docker, and Redis in a professional team environment.
• Deepened understanding of scalable backend architecture and asynchronous task management patterns.