Senior Software Engineer — Agentic AI Systems at PureLogics

Architected and deployed multi-agent orchestration pipelines using LangGraph and AutoGen, enabling autonomous task planning and execution across enterprise client workflows. Built tool-use and function-calling agents integrated with external APIs including CRMs, calendars, and communication platforms, reducing manual task overhead by 40%. Designed RAG pipelines with Pinecone and Weaviate for knowledge retrieval, improving AI-powered enterprise assistants' relevance. Implemented guardrails and safety systems including adversarial input detection, tool-call sandboxing, and human escalation triggers for high-stakes decisions. Led LLM engineering training for 12+ engineers and established internal standards for prompt engineering, agent design patterns, and evaluation methodology.

Senior Software Engineer — AI-Integrated Enterprise Systems at Azets

Designed and deployed AZ SF AB 01 — a Claude Opus-powered AI code assistant for Azets internal engineering teams, enabling autonomous code review, bug detection, and refactoring suggestions across the enterprise codebase. Implemented agentic tool-use pipelines allowing the assistant to navigate codebases, run tests, and propose multi-file edits with structured reasoning traces. Designed safety guardrails and human-escalation fallback mechanisms ensuring reliable, auditable behaviour in a regulated accounting software environment. Integrated LLM-powered document parsing and workflow automation into enterprise mobile applications serving accounting clients across Europe.

AI Voice Agent Engineer (Contract) at Booya AI

Built end-to-end autonomous voice agents capable of conducting natural-language phone calls, extracting structured information, and updating records in Salesforce and HubSpot without human intervention. Applied RLHF-informed reward modelling to optimise agent conversation strategies, improving task completion rates across diverse call scenarios. Designed conversation state machines and fallback strategies handling adversarial inputs, off-topic digressions, and ambiguous user intent in production.

Software Engineer — AI/ML Health-Tech Platform at MarkiTech.AI

Contributed to SenSights, an AI-powered Remote Patient Monitoring (RPM) platform integrating FDA-approved IoT devices, CGM sensors, and wearables. Implemented real-time ML inference pipelines for diabetes and hypertension monitoring, processing continuous biometric data streams with sub-second latency. Built human-in-the-loop workflows for clinical decision support, structuring AI-generated health alerts with clinician override mechanisms and audit logging for regulated deployment.