Arash Khoshbakht
Software Developer
19+ years exp
5+ projects
TypeScript
C#
JavaScript
SQL
Python
Experimentation & Evaluation: 4/10
Active 8 days ago
+90 (534) 3693930 Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
A senior-level backend engineer focused on reliable, auditable transaction and connector systems with strong attention to durable execution and safety for agent-initiated actions. The strongest proven skill is resilient Postgres-backed execution and schema evolution demonstrated by persistence_postgres models, multiple alembic migrations and comprehensive reliability tests (tests/reliability/test_phase3_execution.py and packages/persistence_postgres/*). There is limited public evidence of large-scale operational load-testing, CDN or multi-region distributed systems responsibility and some UI/frontend work exists but not as a primary backend mastery signal.
Technical skills
TypeScript
C#• Junior • 19y+
JavaScript• Middle • 19y+
SQL• Senior • 19y+
Python• Senior • 10y+ • 5+ projects
C#
Entity Framework Core
MassTransit
ASP.NET Core• 5y+
Python
Asyncio
Databases
RabbitMQ• 5y+
PostgreSQL
AI/ML
LangChain
Langfuse
LangGraph
LlamaIndex
OpenRouter
Streamlit
LLM
Prompt Engineering
RAG
AI Agents• 3 projects
Frontend
Bootstrap
React.js
Zustand
DevOps
Azure
GitHub Actions
GitLab CI
Rest API• 19y+
CI/CD• 9y+
Azure DevOps• 8y+
Docker• 5y+
Kubernetes• 5y+
AWS
Mobile
Clean Architecture
QA
Sentry
Timeline
Senior AI Engineer
•
Senior
Vendoroo
•
Full-Time
Designed and maintained production AI and backend workflows for property-management operations. Built LLM-powered agent workflows and asynchronous processes for resident support and operational triage, exposing functionality through REST APIs. Integrated Python services with internal and third-party platforms plus messaging via RabbitMQ and improved AI reliability using prompt/context techniques, evaluation, and regression testing. Enhanced observability and incident handling with Sentry, structured logs, and Slack alerts.
Python
FastAPI
PostgreSQL
SQLAlchemy
Docker
RabbitMQ
Rest API
LLM
RAG
Prompt Engineering
Sentry
Slack
Senior Software Engineer
•
Senior
P&T Solutions
•
Full-Time
Developed and maintained backend services for terminal, port, shipping, cargo, billing, and invoicing workflows. Worked on microservice-oriented systems using .NET Core with SQL, Docker, Kubernetes, and message brokers, focusing on reliability and performance. Contributed to CI/CD and release processes, supporting build, test, packaging, and deployment workflows across environments. Used Azure DevOps release pipeline concepts to enable controlled deployments.
SQL
Dockersince 2021
Kubernetes
Azure DevOps
CI/CD
ASP.NET Core
Python Developer
•
Middle
Kozmos
•
Full-Time
Designed and developed a Django-based website and supporting web APIs with an emphasis on scalable and maintainable architecture. Integrated third-party services and delivered backend improvements through iterative development and testing. Collaborated closely with stakeholders to refine requirements and improve user-facing functionality.
Python
Django
Rest API
Python Developer
•
Middle
Depna
•
Full-Time
Built financial analysis tools and trading automation systems using Python. Applied backtesting-oriented concepts and algorithmic decision logic to support data-driven workflows. Implemented testing and optimization practices to improve reliability of analytics and automation pipelines.
Python
Senior Software Engineer
•
Senior
Amn Pardazan Kavir
•
Full-Time
Led and coached engineering team members, coordinating priorities while reviewing implementation quality to support delivery. Designed and implemented CI/CD pipelines using TFS and Azure DevOps to automate builds, testing, and release workflows. Managed Azure DevOps repositories, build and release pipelines, branching strategies, and deployment troubleshooting across development, staging, and production. Contributed to backend and full-stack development, database design, automated testing, and ongoing production support.
Azure DevOpssince 2017
CI/CDsince 2017
Python & .NET Developer
•
Middle
Atrin Electric
•
Full-Time
Developed scalable web-crawling and data-extraction solutions using Python-based automation workflows. Used BeautifulSoup and Scrapy to improve extraction efficiency and data accuracy for internal reporting. Worked with teams to translate requirements into reliable tooling and improved data processing outcomes over time.
Pythonsince 2018
Beautiful Soup
Scrapy
Full Stack / .NET Developer
•
Middle
Rayatis, Nashr IT, Afkar Sazan Kavir
•
Full-Time
Delivered web applications and business systems using .NET with REST/SOAP interfaces and structured testing practices. Worked with customers and end users to gather requirements, resolve production issues, and produce maintainable software solutions. Implemented frontend functionality using JavaScript and basic web UI technologies and supported ongoing maintenance in production.
JavaScriptsince 2007
Rest APIsince 2007
Full Stack / .NET Developer
•
Middle
Rayatis
•
Full-Time
Delivered web applications and business systems using .NET with SQL Server alongside JavaScript and HTML/CSS. Implemented and integrated REST/SOAP APIs and applied structured testing practices. Worked with customers and end users to gather requirements, resolve production issues, and produce maintainable software solutions.
C#
SQLsince 2007
JavaScript
Rest API
Imam Reza University
Master's Degree •
Artificial Intelligence
Junior AI/ML Engineer
Confidence: Medium LLM Engineer
LLM engineer with senior-level systems engineering skills focused on building transaction-safe agent workflows and production FastAPI services. The strongest proven skill is designing durable, idempotent execution and persistence layers as shown by packages/persistence_postgres/* and packages/application/execution/engine.py. There is little to no evidence of custom model training, ML research, quantization, or GPU optimization in the public code.
Model Architecture & Training
1/10
How well models are designed and trained
Almost no model building or training artifacts; code uses hosted LLM clients rather than custom architectures or training loops.
Evidence
auto-triage-agent/app/core/llm.py: get_chat_model
Data Pipeline & Feature Engineering
1/10
How data is prepared for models
Minimal data-pipeline work; the code collects logs and metrics for LLM inputs but does not contain serious ETL/augmentation or feature engineering.
Evidence
auto-triage-agent/app/agent/nodes.py: collect_context_node (calls provider.search_logs and gathers endpoint metrics)
Experimentation & Evaluation
4/10
How results are measured and tested
Concrete, repeatable testing and benchmarking are present with unit, integration and end-to-end tests plus scenario benchmark code, indicating a disciplined evaluation and acceptance workflow.
Evidence
tests/e2e/property_operations/test_phase6_benchmark.py
demo/property_operations/scenarios/benchmark.py
tests/integration/test_phase3_api.py
MLOps & Deployment
3/10
How models are shipped to production
Production-oriented deployment artifacts and patterns exist (FastAPI app, uvicorn entry, database migrations and repository abstractions) but there is no specialized ML serving or drift-monitoring pipeline.
Evidence
apps/api/app.py: FastAPI application factory and middleware wiring
pyproject.toml: uvicorn and alembic declared in dependencies
packages/persistence_postgres/migrations/env.py: Alembic migration harness
Computational Efficiency
2/10
How efficiently computing resources are used
Some attention to latency and benchmark measurement in scenario code, but no GPU/quantization/packing optimizations or low-level compute engineering.
Evidence
demo/property_operations/scenario/phase2_demo.py: uses perf_counter_ns and computes latency metrics
demo/property_operations/scenarios/benchmark.py: produces latency and benchmark summaries
Research Depth & Innovation
3/10
Depth of research and new ideas
Solid systems-level innovation around durable, idempotent execution, canonical digests and receipt integrity, but not ML research or novel model algorithms.
Evidence
packages/application/receipt/integrity.py: receipt digest and hash-chain validation
packages/core/canonical.py: canonicalization and digest helpers
packages/persistence_postgres/worker_store.py: durable worker lease and reconciliation logic
Expertise
AI Agents & Agentic Workflows• Junior
LLM• Junior
Industries
Artificial Intelligence• Junior
Software• Junior
Technologies
SQL• Senior • 19y+
PostgreSQL
LangGraph
Rest API• 19y+
LangChain
LlamaIndex
Azure DevOps• 8y+
GitHub Actions
Prompt Engineering
GitLab CI
Langfuse
Azure
CI/CD• 9y+
AWS
Docker• 5y+
Kubernetes• 5y+
LLM
RAG
Streamlit
OpenRouter
Recommendations
- Develop production agent orchestration and safety infrastructure that requires durable execution, idempotency and human-in-loop approvals.
- Build connector SDKs and verified adapters for external systems that need strong reconciliation and compensation semantics.
- Implement end-to-end integration and acceptance test harnesses for complex, stateful workflows including database migrations and reproducible benchmarks.
- Lead integrations of LLM-based workflows into existing backend services where strict auditability and non-destructive guarantees are required.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Junior Backend Developer
Confidence: Medium API Engineer
A senior-level backend engineer focused on reliable, auditable transaction and connector systems with strong attention to durable execution and safety for agent-initiated actions. The strongest proven skill is resilient Postgres-backed execution and schema evolution demonstrated by persistence_postgres models, multiple alembic migrations and comprehensive reliability tests (tests/reliability/test_phase3_execution.py and packages/persistence_postgres/*). There is limited public evidence of large-scale operational load-testing, CDN or multi-region distributed systems responsibility and some UI/frontend work exists but not as a primary backend mastery signal.
API Design
6/10
How well APIs are designed
API surface is deliberately designed with a typed async Python client, explicit connector contracts and contract-focused tests that emphasize stable operation ids and OpenAPI compatibility; some API factories are generic scaffolding but explicit tests and a pinned client show careful API ergonomics and backward-compatibility thinking.
Evidence
packages/sdk_python/client.py: CommitLayerClient and TransactionsResource client implementation
packages/connector_sdk/contracts.py: ConnectorManifest and related manifest/contract classes
tests/contract/test_public_contracts.py: tests asserting stable openapi operation ids and registry contract behavior
Data Layer & Database
7/10
Working with databases
Strong, production-aware relational data layer with Alembic migration history, explicit SQLAlchemy models, well-separated repository layer and tests for migrations and RLS; transaction conflict handling and explicit persistence primitives are implemented.
Evidence
packages/persistence_postgres/models.py: Declarative SQLAlchemy models defining domain tables
packages/persistence_postgres/migrations/versions/02d7b3ea58a5_phase_3_durable_execution_baseline.py: alembic migration
tests/migration/test_phase3_migration.py: test validating migration upgrade and drift
Scalability & Performance
6/10
Handling load and speed
Scalability patterns are present via a leasing worker model, durable worker store and explicit retry/lease behavior; there is less direct evidence of broad caching, distributed rate limiting, or measured load-test artifacts.
Evidence
packages/application/execution/engine.py: worker leasing and run_worker_once/WorkerCrash handling
packages/persistence_postgres/worker_store.py: lease_next and scheduling functions
demo/property_operations/scenarios/benchmark.py: scenario benchmark and latency/metrics collection
System Architecture
7/10
Overall system structure
Clear modular architecture separating core, connectors, persistence and SDK layers, with docs and scripts enforcing package boundaries and ADRs; decomposition is purposeful rather than microservices-for-show.
Evidence
scripts/check_architecture.py: architecture enforcement tooling
docs/ARCHITECTURE.md: architectural rationale and phase descriptions
packages/connector_sdk/registry.py: explicit runtime registry and handler dispatch design
Security & Auth
7/10
Protecting data and access
Security and auth are treated as first-class concerns with an OIDC validator/middleware, scoped callback HMAC authentication, secret manager abstraction and RLS-focused migrations and tests enforcing tenant safety.
Evidence
packages/auth_oidc/middleware.py: IdentityMiddleware and header replacement logic
packages/secrets/callbacks.py: HmacCallbackAuthenticator and scoped callback secret resolution
packages/persistence_postgres/migrations/versions/7b6f2e8a91c4_phase_8_tenant_security.py: migration adding tenant security/RLS
Reliability & Observability
7/10
Stability and monitoring
Reliability patterns are well addressed: immutable domain entities, receipt hash-chain integrity, extensive tests for crash/retry/lease semantics and verification/compensation flows to survive partial failures.
Evidence
packages/application/receipt/integrity.py: receipt digest and hash-chain validation
tests/reliability/test_phase3_execution.py: many tests around timeout, restart, fencing and reconciliation
packages/application/compensation/engine.py: compensation lease and run logic
Expertise
Backend AI & LLM• Junior
Databases & Vector Storage• Junior
Microservices & API Architecture• Junior
Python• Junior
Industries
Artificial Intelligence• Junior
Information Technology• Junior
Technologies
Asyncio
Recommendations
- Develop durable transactional integration platforms that require idempotency, compensation and verification logic (connector and durable-repository work).
- Own Postgres-backed backend services with strict tenant isolation and migration chains including RLS and migration tests.
- Build API-first SDKs and contract-driven connector registries or public REST/async clients for partner integrations.
- Implement safety-critical agent orchestration or LLM-assisted workflows where human-in-the-loop approval and auditability are required.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Intern Frontend Developer
Confidence: Low App Engineer
Junior frontend developer focused on React-based single page applications with an emphasis on building UI-driven functionality. The strongest proven artifact is a package.json showing a React-based frontend using Zustand for state (evidence: frontend/package.json dependency list). There is no original, high-ownership human-authored frontend component code, state-machine logic, performance tuning artifacts, or test suites visible in the public human-authored files to support senior-level claims.
UI Component Architecture
How interface parts are built
Not evidenced in public code
Responsive & Cross-browser
Works on all screens and browsers
Not evidenced in public code
Performance Optimization
Speed of the interface
Not evidenced in public code
Accessibility & Semantics
Usable for everyone
Not evidenced in public code
State Management & Data Flow
Managing data in the app
Not evidenced in public code
UX & Visual Polish
Look and feel quality
Not evidenced in public code
Expertise
React• Intern
Technologies
Zustand
Bootstrap
React.js
Recommendations
- Assign to implement or extend small React features and components using Zustand for local state under senior mentorship.
- Use for building or iterating on visual UIs such as node-based editors or toolbars where React knowledge suffices and scope is well defined.
- Have the developer add tests, explicit async-state handling (cancellation and optimistic updates), and accessibility improvements to existing components to grow reliability.
- Pair with a senior engineer to refactor and own one UI area end-to-end, including CI linting, a11y checks, and storybook documentation.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: