Jonathan Bogdanovicz

Brings 20+ years of software engineering and 4+ years of AI/ML experience to a Senior AI/ML Engineer role focused on NLP, LLM, and document-understanding systems for regulated healthcare and insurance-adjacent enterprise workflows.

Designs

Python and AWS Bedrock solutions that combine RAG, classification, extraction, AI agents, and evaluation harnesses to turn unstructured documents into auditable business actions. Integrates AI services with React, Node.js, MongoDB, and cloud data platforms, using secure APIs, tenant isolation, and observability to support production delivery. Communicates complex tradeoffs with distributed teams and stakeholders across time zones, translating ambiguous requirements into reliable, measurable systems.

Senior AI/ML Engineer at Yeda Health (2025-12 – Present)

• Built Python and LangChain services that classified clinical documents and extracted structured facts, reducing manual review time by 38% across intake queues.
• Engineered retrieval-augmented generation workflows on AWS Bedrock, grounding answers in consent-scoped records and enforcing citation checks before routine clinician review.
• Designed tool-calling AI agents that coordinated data lookups, meal-photo analysis, and follow-up recommendations through guarded backend APIs and audit trails.
• Integrated React Native and React interfaces with Python inference services, improving response visibility and reducing failed user sessions by 24% across partner pilots.
• Partnered with clinicians and data scientists to define evaluation sets, privacy rules, and safe fallback behavior for LLM-generated insights before production releases.

Founder & Lead Software Engineer at Solutio.Me (2007-01 – 2026-09)

• Modernized an Elasticsearch and MongoDB product-search stack to PostgreSQL and vector retrieval, reducing complex query latency by roughly 70% for large catalogs.
• Implemented Python microservices for document classification, entity extraction, and semantic search, supporting four AI-assisted products across billing workflows over four years.
• Led migration of AI-assisted billing workflows to event-driven AWS services, increasing payment recovery by 17% while materially lowering delinquency risk.
• Established architecture reviews and secure API standards for a multi-tenant platform, guiding six engineers through production releases and incident response.
• Automated evaluation and regression checks for retrieval quality and classification accuracy, reducing prompt regressions discovered after deployment through golden datasets and release gates.

Co-Founder & CTO at Mendy (2024-07 – 2025-06)

• Built document question-answering and classification services with Python, LangChain, and AWS Bedrock, improving extraction accuracy by 29% on partner datasets.
• Developed retrieval and tool-calling agents that automated approved ERP, CRM, and web workflows, querying financial data through least-privilege connectors and tenant-scoped policies.
• Architected tenant authorization, prompt-context isolation, and audit logging for AI requests, strengthening data separation across partner organizations and downstream customers.
• Improved conversational response latency by 41% through streaming APIs, asynchronous orchestration, and targeted caching of approved metric definitions across high-volume pilot workloads.
• Mentored a compact engineering team on LLM evaluation, prompt versioning, and release rollback, raising pilot readiness before partner-facing enterprise demos.

Senior Systems Analyst at COAMO Agroindustrial Cooperativa (2014-07 – 2019-11)

• Developed an online banking-style member portal with Node.js, JavaScript, and Ionic, enabling remote crop sales and loan access across branches.
• Integrated SOAP services and Uniface workflows, synchronizing financial and agricultural records and reducing manual reconciliation errors by 22% during harvest peaks.
• Designed a Fideliza points ledger with idempotent accrual, reversal controls, and audit trails, preserving member balance accuracy across seasonal purchase cycles.
• Optimized transactional workflows for grain receipt and inventory updates, improving batch throughput by 1.8x during peak harvest periods across receiving units.
• Supported production releases and mentoring for business-critical applications, documenting rollback steps and troubleshooting routines for support teams under seasonal operating constraints.

PHP Developer at Google MKT (2010-07 – 2013-07)

• Built responsive product catalogs, dynamic filters, and pricing variants with PHP and jQuery, supporting seasonal promotions without manual catalog rebuilds.
• Integrated REST and SOAP APIs with ERP systems, synchronizing inventory, pricing, and customer data and reducing stock discrepancies by 18%.
• Implemented back-office order management and campaign reporting features used daily by sales and operations teams across multiple regional client accounts.
• Collaborated with designers, account managers, and clients to translate campaign requirements into maintainable web applications, test plans, and release checklists.

PHP Developer at Alogeniun (2005-08 – 2010-06)

• Created custom PHP and JavaScript websites with dynamic interactions, improving client usability across major desktop browsers and selected legacy devices.
• Connected web applications to ERP systems through SOAP services, synchronizing catalogs, inventory, and customer records in real time for mid-market clients.
• Tested, debugged, and optimized application performance, documenting deployment steps and supporting post-release client handovers and technical acceptance alongside senior developers.

Bachelor's Degree in Information Systems – Centro Universitário UDC (2008 – 2012)