2+
Years of shipping
SJ
Suhaib Jbara
AI / Software Engineer
AI + Cloud Engineering
AI and cloud engineer who builds real products and carries them to production.
I'm Suhaib Jbara, an AI-first software engineer and cloud engineer based in Lebanon. I build agentic AI applications, RAG systems, full-stack products, backend APIs, mobile features, observability workflows, network-aware systems, and cloud deployments for teams that need someone who can turn a business problem into working software.
AI + Cloud
Core domains
AUB CS '24
Education
ProfileAvailable
Suhaib Jbara
AI / Software Engineer
Lebanon · Open to opportunities
AI, cloud, full-stackResume
Professional timeline
Full Stack Software Engineer
Aligator Technology LLC
Dec 2024 - Aug 2025
Integrated external services such as Reddit and SendGrid to enrich product capabilities and user communication flows.
Worked on landing-page delivery and deployment as part of the company's branding and outreach efforts.
Built threshold-based alerting and an internal newsletter feature to improve visibility and team coordination.
Created an internal observability system with Grafana Cloud to track system health and support incident response.
Software Engineer Intern
Aligator Technology LLC
Jun 2024 - Aug 2024
Cleaned and filtered raw scraped JSON data so downstream product views used higher-quality information.
Optimized database and ORM queries, contributing to a reported 50% performance improvement.
Software Engineer Intern
Pronto LLC
May 2024 - Jul 2024
Delivered new features in a React Native mobile application and improved signup and profile-update validation.
Enhanced search with fuzzy search, popular search, and related-product improvements.
Implemented a first-order reward flow to improve early user engagement.
ResumePDF
Optimized for your records.
While this website provides an interactive deep-dive into my work, my resume is the condensed, ATS-ready summary of my engineering journey. Download it for a focused version optimized for recruitment pipelines.
2+ yrs
Experience
Full
Focus
Format
Download ResumeATS-Optimized · Updated September 2026
How I work
01
Strong software engineering fundamentals, grounded in computer science and applied through real product delivery.
02
AI-first mindset with a strong focus on building AI-powered applications that solve real-world problems.
03
Architecture that balances product speed with operational clarity, not just feature output.
04
Cloud and deployment ownership treated as part of the engineering scope, not separate from development.
05
Interfaces built to feel polished while still serving real workflows and business problems.
Featured Projects
View all projectsSelected work
A curated selection of projects that represent how I think about engineering — from product surface to deployment pipeline.
fullstackcloud
Tagushi
Tagushii is a production-grade sushi restaurant ordering platform built around a realistic local-business use case. What makes it worth highlighting is the full-stack ownership — from the Angular SPA surface to the Cloudflare Worker API, D1 database with Drizzle ORM, admin panel with CRUD management, real-time Telegram notifications, and slot-based availability scheduling. Every piece — frontend, API, database, CI/CD, and cloud infrastructure — was built and shipped as a single engineering unit.
AngularCloudflare WorkersD1Drizzle ORMREST APIsAdmin DashboardOrder ManagementAvailabilityOffersTelegramCart & CheckoutGitHub ActionsCI/CD
A complete online ordering system built for a sushi restaurant — customers browse the full menu, add items to their cart, pick a delivery time slot with real-time availability, and place their order in seconds. The experience is fast, works on any device, and requires no account to use.
The restaurant owner manages everything from a single dashboard — update menu items, track and update orders, set opening hours, and create time-limited offers and discounts. Customers always see current pricing, real-time slot availability, and active promotions without ever needing to refresh.
Deployed on Cloudflare's global edge network with zero server maintenance and zero ongoing cost — the site loads fast anywhere in the world. Every new order is instantly pushed to the restaurant via Telegram so nothing gets missed.
The entire system — website, API, database, and deployment — is defined in code and ships automatically through GitHub Actions whenever changes are pushed. No manual steps, no configuration drift.
aiclientConfidential
AI Customer Service Agent
Built for Zetheta Algorithms Private Limited
Confidential client work. A production-style AI customer-service platform for a digital bank handling tens of thousands of daily interactions. The system understands intent and sentiment, answers from a continuously updated knowledge base, enforces strict safety and compliance boundaries, routes critical cases to human experts, and learns from supervisor feedback — all surfaced through real-time operator dashboards with no fabricated metrics.
Agentic AIAI SupportSafety GuardrailsHuman-in-the-LoopRAGReal-time Analytics
Built conversational understanding (intent, entities, sentiment) with escalation routing and supervisor-feedback learning loops.
Operated on a strict principle of no fabricated data — every dashboard metric computed from real pipeline runs.
Live operator dashboards with real-time, daily, and weekly views for human oversight.
Business impact
Engineered to automate ~60% of first-line support and cut first-response time from ~2 minutes to under 5 seconds.
Designed to raise containment from ~35% to 70%+ while holding incorrect financial advice at 0%.
Validated: adversarial attempts (injection, jailbreak) blocked at 0.94–0.97 confidence across documented runs.
500+ automated tests across the safety and support pipeline.
aisystemsclientConfidential
Compliance Monitoring System
Built for Zetheta Algorithms Private Limited
Confidential client work. A compliance-monitoring platform for a financial institution that orchestrates specialized AI agents to surveil trading, lending, and client-communications activity. It flags misconduct such as market abuse, AML structuring, and sanctions exposure, routes uncertain cases to human experts through a structured escalation workflow, and maintains a tamper-evident, jurisdiction-aware audit trail.
Agentic AIRisk & ComplianceReal-time MonitoringHuman-in-the-LoopObservability
Multi-agent surveillance with probabilistic consensus reconciliation and tiered, role-based human escalation.
Tamper-evident audit trail scoped to jurisdictional retention requirements.
Observability dashboards and a load-test harness built for production monitoring.
Business impact
100% coverage across 20 regulatory scenarios (incl. insider trading, AML structuring, sanctions evasion).
Zero false alerts on a legitimate high-volume trade case — target false-positive rate under 5%.
Sub-500ms detection on trade alerts; coordinated multi-signal cases flagged in under 2 minutes.
Designed to handle ~2.4M transactions and ~850K communications daily across 23 regulatory bodies.
cloudsystems
Internal Observability Platform
Designed and implemented an internal observability platform using Grafana Cloud to track system health metrics, support incident response workflows, and improve engineering team visibility into production systems. Included threshold-based alerting and an internal newsletter feature.
Grafana CloudObservabilityAlertingDevOpsMonitoring
Built threshold-based alerting system to proactively detect and respond to production issues.
Created dashboards for real-time system health monitoring across multiple services.
Integrated with existing deployment pipelines for end-to-end operational awareness.
aicloudfullstack
JobCrew — Multi-Agent Application Platform
JobCrew replaces the generic chatbot approach to job applications with a transparent, coordinated crew of AI agents. Users submit a job posting and their resume, and a LangGraph-powered workflow of specialized agents executes in sequence — each researching, validating, profiling, and generating tailored artifacts. The entire system runs on AWS Bedrock for LLM inference and semantic search, deployed via Terraform to cloud infrastructure, secured behind Cloudflare Tunnel, and continuously delivered through an automated CI/CD pipeline with built-in security guardrails, rate limiting, and audit logging.
AWS BedrockLangGraphTerraformCI/CDCloudflare TunnelFastAPINext.jsChromaDBRAGAgentic AI
Orchestrated five specialized AI agents (Researcher, Validator, Profiler, Resume Strategist, Interview Preparer) using LangGraph — agents execute in a dependency-aware directed graph with real-time WebSocket and SSE streaming so users watch the workflow unfold.
Built entirely on AWS Bedrock for both LLM inference and semantic embeddings — enabling grounded, hallucination-resistant resume tailoring through a ChromaDB vector store with RAG retrieval.
Full infrastructure defined as code with Terraform — single apply provisions all cloud resources including container registries, storage, IAM roles with least-privilege Bedrock permissions, secret management, and centralized logging.
Zero-trust networking with Cloudflare Tunnel — no public ports exposed on the server; all traffic routes through Cloudflare's edge with automatic HTTPS and DDoS protection.
Automated CI/CD pipeline via GitHub Actions — Terraform provisions infrastructure, Docker images are built and pushed, then the application deploys remotely with zero manual steps.
Production security guardrails including prompt injection detection, request rate limiting, PII sanitization, input validation, and structured audit logging to CloudWatch for full observability.
aiclientConfidential
WealthPilot AI — Autonomous Portfolio Rebalancing
Built for Zetheta Algorithms Private Limited
Confidential client work. A multi-agent system that continuously monitors client portfolios against target allocations, detects drift, and generates optimized, tax-aware rebalance plans — each validated against risk and regulatory constraints before execution, with human approval paths and an immutable audit trail. Every decision ships with three explanations: plain language for clients, quantitative detail for advisors, and audit-grade records for regulators.
Agentic AIMulti-Agent SystemsRisk & ComplianceExplainabilityHuman-in-the-Loop
Designed and built the multi-agent orchestration pipeline, coordination patterns, failure handling, and human-in-the-loop escalation system.
Produced three audiences of explanations (client / advisor / regulator) with numeric-consistency checks that catch hallucinated figures.
Built a backtesting engine validated across crash, rate-hike, bull, and full-cycle market periods.
Business impact
Simulated backtests (2019–2024) vs legacy calendar rebalancing: ~1/3 lower trade volume and transaction costs, and more than half the average portfolio drift (6.3% → 2.9%).
Improved risk-adjusted returns (Sharpe 0.48 → 0.72) including better downside behavior during the 2020 crash.
Scales to 50,000 portfolios with sub-30-second full scans; 248 automated tests.
Compliance scorecard + human-override workflow make every decision auditable.
aifullstack
Notion Intelligence Hub
An AI-driven research tool built with Next.js and FastAPI. It features a full RAG (Retrieval-Augmented Generation) stack that indexes Notion pages and attached notebooks into a local ChromaDB instance, enabling high-precision retrieval with both 'Fast' and 'Agentic' reasoning modes.
Next.jsFastAPILlamaIndexChromaDBOllamaAgentic AIWebSockets
Engineered an Agentic 'Thinking Mode' using LlamaIndex Workflows for multi-step reasoning and autonomous search.
Developed a custom ingestion pipeline for Notion blocks and .ipynb notebooks with MD5-based delta indexing.
Implemented a real-time streaming architecture using WebSockets to visualize the agent's internal reasoning process.
Designed for privacy-first use cases, supporting 100% local inference via Ollama and persistent ChromaDB storage.
fullstackcloud
Portfolio Website
A personal portfolio built with Next.js 16, featuring an interactive particle canvas background, pointer-reactive 3D card tilt effects, glassmorphism UI elements, and a fully data-driven projects system. Deployed on Cloudflare with GitHub Actions CI/CD.
Next.jsTypeScriptCanvas APICloudflareCI/CD
Custom particle canvas animation with dynamic connection lines between nearby particles.
Pointer-reactive 3D perspective tilt on the engineer profile card.
Data-driven project management — add projects by editing a single TypeScript file.
Deployed on Cloudflare edge network with automated GitHub Actions pipeline.
Engineering Philosophy
Build the AI product. Own the cloud pipeline.
Every project I ship follows the same arc: understand the stakeholder problem, build the application, automate delivery, add observability, then own the production outcome.
Step 01
Problem Modeling
Start from a real business need, then shape the system around delivery, reliability, and user flow rather than code alone.
Step 02
Application Build
Implement the product surface with maintainable frontend and backend patterns, clear state flow, and real user interactions.
Step 03
Cloud Delivery
Move beyond local success into repeatable build, deployment, and hosting workflows with environment-aware automation.
Step 04
Operational Thinking
Treat observability, reliability, and release flow as part of engineering ownership instead of afterthoughts.
Technical stack
Languages
Python, TypeScript, JavaScript, Java, Kotlin, SQL, C/C++, C#
AI
Generative AI, LLMs, RAG, Agents, LangChain, Vector Databases
Frontend + Mobile
NextJS, React, Angular, React Native, Jetpack Compose
Backend + APIs
Node.js, FastAPI, Spring Boot, Rails, .NET, REST APIs
Cloud + Ops
AWS, Cloudflare, Docker, Terraform, Grafana Cloud, GitHub Actions, CI/CD
Network + Automation
MikroTik, CAPsMAN, RADIUS, VPN, MQL5, Risk Automation
Academic foundation
American University of Beirut
B.Sc. Computer Science | Class of 2024 | GPA 3.69/4.0
USAID Higher Education Scholarship recipient and Dean's Honor List distinction.
Carnegie Mellon University in Qatar
Exchange Semester | Spring 2024
Community and leadership
Community-based internship teacher, IEEE Student Branch webmaster, Computer Science Student Society member at large, and 100+ volunteer hours.
Credentials
Contact
Looking for someone who can build the app and think through the deployment path too?
I'm available for AI engineering, cloud engineering, and full-stack software roles where product delivery, system quality, stakeholder clarity, and cloud execution all matter.