Danya K.

Project overview:

AI-assisted Sales Pipeline and Lead Generation automation for an event agency. The project replaced fragmented manual tracking across social media, forums, spreadsheets, and an outdated CRM with a structured qualification pipeline, messenger notifications, and automated lead-routing flows.

Responsibilities:

• Designed the end-to-end lead flow from raw message to qualified follow-up
• Developed Excel-to-CRM automation; built provider-routing and fallback logic
• Configured messenger notifications
• Supported local deployment and setup for users through shell and PowerShell scripts

Achievements:

Reduced repetitive copy-paste and duplicate checking across Excel/CRM sources; Introduced AI-assisted lead qualification based on incoming message context; connected social/forum messages to scenario-based follow-up; Enabled messenger alerts for high-priority leads; Implemented fallback logic across LLM providers to improve reliability when quotas or providers were unavailable.

Technology stack:

Project overview:

Custom multi-agent AI development framework for assisted software delivery. The system coordinates specialized agents for architecture planning, implementation, debugging, testing, review, static analysis, and project execution, enabling faster movement from product idea to working demo.

Responsibilities:

• Designed agent architecture, routing rules, and task delegation
• Created workflows for code generation, test execution, static analysis, and review
• Integrated agents with local development context
• Defined guardrails for safer automation
• Refined the framework through real product demo development

Achievements:

Built a multi-agent architecture for coding, testing, and review workflows Added agent roles for implementation, investigation, verification, and orchestration Supported product demo creation from scratch Reduced repetitive engineering work Enabled long-running assisted development sessions while keeping human approval and review steps in the loop

Technology stack:

Project overview:

Legal and procurement AI assistant for document-heavy workflows involving legislative data, tender documentation, document packages, and structured request optimization. The assistant helps users retrieve relevant source material and prepare grounded answers or drafts for review.

Responsibilities:

• Designed retrieval and generation workflows
• Structured document packages for search and analysis
• Built request optimization logic
• Defined assistant behavior for tender analysis, legal references, and procurement support
• Balanced answer quality, cost, and traceability

Achievements:

Integrated retrieval workflows over domain-specific legal/procurement content Used RAG to ground answers in source material Applied DSPy-style optimization to improve prompt quality and repeatable request handling Improved token efficiency for document-heavy workflows Supported traceable AI assistance rather than generic chatbot behavior

Technology stack:

Project overview:

Agent-to-Agent service monitoring and assisted remediation workflow. A cloud watcher checks service health, reports issues, and coordinates with a coding/remediation agent that analyzes severity, prepares a report, and sends it to messenger for human approval before changes are applied.

Responsibilities:

• Designed watcher-to-coder agent orchestration
• Implemented health checks, status reporting, and escalation logic
• Defined safety boundaries for automated operations and provider-management layers

Achievements:

Automated error detection and reporting; Connected monitoring agents with remediation agents; Added messenger approval flows; Improved response speed for service issues; Kept fixes controlled through human-in-the-loop; Approval rather than uncontrolled automation;

Technology stack: