About Me

Lead the backend for an insurtech mobile platform that aggregates multiple insurance providers behind a single unified experience.

• Unified 5 insurers—each with its own API—behind a single interface so users quote and buy policies in one place; the provider-agnostic design lets the business onboard new insurers without rewriting the platform (scalable to 10+).
• Built quoting, underwriting and policy issuance flows with Django REST Framework, PostgreSQL and Redis, optimizing performance through caching.
• Implemented automated document OCR with AWS Textract and a multi-vendor pharmacy marketplace module.

Built an end-to-end production AI voice phone assistant serving 10 multi-tenant clients and 5,000+ call minutes per month.

• Orchestrated a real-time voice pipeline with Pipecat (Deepgram STT → GPT-4.1-mini → Cartesia TTS) achieving sub-800ms latency, with barge-in, end-of-turn detection and VAD across PSTN ↔ WebRTC ↔ WebSocket.
• Built a dynamic conversational flow engine driven by JSON/API—new call flows ship without code changes—with episodic memory, per-client identity and prompt-injection protection.
• Developed a telephony broker orchestrating a headless Chromium (Playwright) running RingCentral's SIP/WebRTC WebPhone, with a custom PCM/WAV frame serializer and real-time observability.

Backend and AI architecture for high-volume SaaS products under NDA.

• Architected the backend of a high-volume automated audio transcription platform (Madrid), orchestrating jobs with async task queues and a privacy-first zero-knowledge authentication mechanism.
• Architected AI agents for a voice-based recruitment platform (Miami) that ranks talent and delivers voice summaries, with a classification engine over 6,000+ metadata-rich records.
• Built asynchronous batch-processing architectures for continuous parallel workloads and established system observability standards.

Led the architecture and team of a business-critical multi-service platform on AWS, after delivering its core automation as a full-stack developer.

• Delivered an automated property valuation system processing 5,000+ listings per cycle, cutting valuation time by 70% and reaching up to 90% pricing accuracy.
• Led a team of 2 developers, established engineering standards (security, observability, deployments) and implemented CI/CD pipelines with Jenkins.
• Built data extraction pipelines (Scrapy, BeautifulSoup) and comparative pricing models with Pandas, plus interactive dashboards for internal teams.

Designed and deployed complex automations for business-critical B2B workflows.

• Built n8n automations integrating chatbots, LLM pipelines and SaaS APIs, enhanced with Python and JavaScript for event-driven flows.
• Synced data across CRM, ERP and marketing platforms through internal and third-party API integrations, and added AI capabilities (LLM APIs, vector search, prompt engineering).

Designed and optimized automated workflows connecting disparate systems with no-code/low-code tools and API integrations.

• Analyzed business requirements to build custom automation (n8n + API integrations), reducing manual tasks and driving digital transformation across operational units.