Diving deep into complex systems
Tal Zigelnik
Engineered for End-to-End Ownership
Computer Science graduate (B.Sc.) and full-stack/backend engineer with hands-on experience in cloud infrastructure, RESTful integrations, and agentic AI workflows. I translate complex business logic into reliable, production-ready applications.
~50m · Sunlight Zone
Still enough light here to see clearly.
I'm a software and systems engineer who likes owning things end-to-end, from the database schema to the deploy pipeline that ships it. Recent work spans real-time backends, ML-adjacent processing pipelines, and cloud infrastructure for small teams moving fast.
Backend Architecture
REST APIs, service boundaries, async pipelines
Cloud Deployments
CI/CD, Vercel, distributed storage
API Integrations
Third-party services, auth, real-time sync
AI Workflows
Inference pipelines, agentic tooling
~150m · Edge of the Sunlight Zone
Projects
Kitchen Agent
Agentic Telegram bot that turns a Hebrew voice note into a priced PDF quotation.
- Uses the Claude API with structured tool-use to extract a job spec from a Whisper transcription, biased with a custom Hebrew carpentry-trade glossary for accuracy.
- Keeps pricing logic deterministic and separate from the LLM by design, then generates an RTL PDF quote and delivers it back over the Telegram Bot API.
Oosht
Real-time photo sharing during live events via QR code.
- Designed and built a scalable backend using REST APIs and Firebase/Firestore for real-time data synchronization.
- Supported 50+ concurrent users uploading media simultaneously with real-time updates across clients.
FractureCapture
★ 1Web application detecting and localizing bone fractures in X-ray images with YOLOv8.
- Reached 81% fracture detection accuracy with YOLOv8, after leading data gathering and labeling and resolving inconsistent bounding-box labels and polygon annotation formats.
- Led a team of 6, including task planning, code reviews, and integration (Agile).
Pneumonia Detection
Deep learning classifier detecting pneumonia from chest X-rays.
- Built a CNN using transfer learning on InceptionV3, benchmarked against a scikit-learn KNN baseline.
- Evaluated with precision, recall, F1-score, ROC-AUC, and confusion matrices, using augmentation and hyperparameter tuning to improve generalization.
Scrabble Game
★ 1Real-time multiplayer game with a client-server architecture.
- Developed a real-time multiplayer game with a multi-threaded client-server architecture.
- Implemented design patterns (Singleton, Observer) to ensure a modular and maintainable system design.
Phonebook
★ 1Full-stack e-commerce platform with auth, products, and cart.
- Built the backend using Node.js, Express, and MongoDB for scalable and efficient data handling.
- Integrated Google Maps, Facebook, and GNews APIs to enhance functionality and user engagement.
~250m · Twilight Threshold
Open Source
Contributions to projects outside my own dive log.
~400m · Twilight Zone
Experience, Education & Skills
Signals returning from deeper water: less light, sharper contrast.
Dec 2025 - Present
IVR Developer
ITNav
- Built an internal Java/Spring Boot tool that compares and deploys Genesys Cloud configuration across environments, with REST controllers, an adapter per resource type (data tables, integrations, prompts, queues, schedules, wrap-up codes), a diff engine, a dependency resolver with cycle detection, and deployments with rollback support.
- Design, build, and maintain IVR call flows in Genesys Cloud Architect, including menus, routing logic, prompts, and schedules that send callers to the right queues.
- Author, version, and deploy Architect flows as YAML with the Archy CLI, and provision Genesys Cloud configuration using Terraform.
- Configure and maintain queues and routing for enterprise Genesys Cloud Contact Center environments, keeping production IVR stable as requirements change.
- Track and analyze flow and interaction logs to diagnose call-flow, routing, and configuration issues, and troubleshoot REST API, Webhook, and OAuth integrations.
- Develop agentic virtual agents (Genesys Cloud AVA) with Claude, using an MCP-based harness and agent skills to define, build, publish, and test agents.
- Work closely with QA, product managers, and developers to investigate and resolve production issues across development, testing, and production environments.
Education
B.Sc. Computer Science
Apr 2022 - Jul 2024The College of Management Academic Studies
Deep Learning Internship (Academic, Final Year)
2023 - 2024Trained and evaluated computer vision models on medical imaging data.
Contact Center
- Genesys Cloud
- Architect
- Archy
- IVR Flows & Queues
- Genesys AVA
Backend & Cloud
- Spring Boot
- FastAPI
- Node.js
- Terraform
- CI/CD
Languages
- Java
- Python
- TypeScript
- React
AI & ML
- Claude API
- Agentic tool-use
- MCP
- TensorFlow / PyTorch
- Claude Code
~1000m · Abyssal Trench
Let's build something down here.
Open to backend, cloud, and AI-workflow projects. Reach out and I'll get back to you.
Built with Next.js, Tailwind, Lenis & Framer Motion.