Christian – React, Node.js, AWS

An agentic workflow platform that embeds AI agents into controlled business processes instead of relying on standalone chatbots. The first use case supports premium clinics, automating the journey from inbound lead to patient evaluation, follow-up, appointment scheduling, payment coordination, and treatment start.

The platform combines multi-channel AI agents for WhatsApp, phone calls, email, chat, forms, and internal systems, LLM-powered reasoning, LangChain, LangGraph, modular agent architecture, RAG with Pinecone, tool/function calling, workflow automation, human-in-the-loop handoffs, observability, evaluation frameworks, and reusable industry-specific packages.

• Designed and built an agentic workflow automation platform that embeds AI agents into structured business processes.
• Built multi-channel AI agent workflows across WhatsApp, phone calls, email, chat, forms, and internal systems, enabling automated lead qualification, patient follow-ups, appointment coordination, missing information collection, payment coordination, status updates, and escalation handling.
• Implemented both text-based and voice-based conversational flows, including WhatsApp messaging, chat interactions, inbound/outbound phone calls, and voice notes, allowing agents to operate across the communication channels used by the business.
• Worked with modern voice AI technologies such as OpenAI Realtime API / GPT Realtime, OpenAI speech-to-text and text-to-speech models, ElevenLabs voice AI, and Microsoft Azure AI Speech / Voice Live API for real-time conversational voice automation, transcription, and speech generation.
• Implemented LLM-powered agents capable of understanding business context, classifying information, requesting missing data, making bounded decisions, and executing actions through real tools and APIs.
• Used LangChain and LangGraph together with a custom modular architecture to orchestrate multi-step agent workflows.
• Implemented RAG workflows with Pinecone to retrieve customer-specific, business-specific, and workflow-specific context for agent decision-making during text and voice interactions.
• Enabled agents to perform tool/function calling across CRMs, calendars, WhatsApp, email, payment flows, internal APIs, external APIs, forms, spreadsheets, support platforms, financial systems, and customer systems.
• Designed human-in-the-loop workflows for escalations, approvals, exception handling, review steps, fallback scenarios, and evidence tracking.
• Built evaluation workflows using prompt regression tests, LLM-as-judge, golden datasets, traces, and scorecards to improve reliability and reduce regressions in agent behavior.
• Implemented observability using CloudWatch, Sentry, custom logs, and internal dashboards to monitor agent execution, text and voice conversation outcomes, workflow progress, errors, handoffs, and business results.
• Designed reusable agentic workflow components separated from customer-specific implementations and industry-specific business packages.
• Worked on prompt design, agent instructions, text and voice conversation flows, guardrails, workflow state management, escalation policies, and testable AI behavior.

An AI-native parenting app that helps parents through personalized courses, pathways, daily sessions, podcasts, and conversational AI experiences. The product uses multi-agent AI to tailor guidance based on family context, user interactions, lessons, chats, and behavioral patterns, making support more interactive, contextual, and actionable.

• Designed and developed production-grade full-stack features across frontend, backend, cloud infrastructure, and AI-powered product workflows;
• Helped rebuild the product from a traditional parenting app into an AI-native experience centered on personalization and conversational guidance;
• Contributed to the design of a production multi-agent system for context management, personalization, and product-level AI interactions;
• Helped design a Global Context Agent that centralized user interactions, lessons, chats, family context, and product behavior into a personalized family record;
• Transformed the chat experience from a hardcoded onboarding flow into an interactive AI experience integrated across the broader product;
• Built AI-powered workflows that helped parents reflect on lessons, apply parenting strategies, and receive more personalized guidance;
• Architected standardized communication layers that allowed LLMs to securely access and interact with internal product data;
• Designed evaluation frameworks to treat prompts, AI behavior, and agent workflows as testable software artifacts;
• Built and maintained scalable services using TypeScript, Node.js, Next.js, Python, and AWS;
• Collaborated with product, design, and operations teams to translate parenting methodology and business requirements into reliable software systems.

An AI platform for software engineers that uses fine-tuned, code-focused LLMs deployed locally or within private company infrastructure. The platform covered the full LLM lifecycle, from dataset collection and fine-tuning to evaluation, quantization, local inference, and product integration for developer-facing AI features.

• Worked as a founding engineer across the full product and AI stack;
• Built full-stack features using React, Node.js, TypeScript, Python, and cloud-based backend services;
• Collected, cleaned, processed, formatted, and labeled coding datasets for fine-tuning code-focused language models;
• Experimented with LLM fine-tuning, knowledge distillation, quantization, and local inference;
• Built evaluation pipelines to compare model outputs, detect regressions, and reduce issues such as overfitting;
• Worked with QLoRA, GGUF, llama.cpp, and local model execution on GPU hardware;
• Used Hugging Face, SageMaker, and Google Colab for larger model experimentation and training workflows;
• Integrated open-source LLMs into proprietary workflows, reducing dependency on external AI providers;
• Helped evolve static chatbot experiences into task-oriented AI features for real developer workflows;
• Established CI/CD and versioning practices for application code, prompts, datasets, and evaluation scripts.

A scalable web and backend platform for internal and customer-facing systems, built with React, Node.js, Python, PostgreSQL, and AWS. The work focused on improving platform scalability, reducing latency, modernizing legacy components, and supporting high-performance services through cloud-native architecture patterns.

• Led cross-functional engineering teams in the delivery of scalable web applications and backend systems;
• Designed and implemented full-stack features across frontend interfaces, backend APIs, database layers, and cloud services;
• Modernized system architecture by migrating components to AWS serverless and microservices-based patterns;
• Improved scalability and reduced latency through better service separation, backend optimization, and cloud-native design;
• Designed and optimized PostgreSQL databases for high-performance, real-time data retrieval;
• Supported engineering quality through code reviews, technical planning, mentoring, and agile delivery practices;
• Collaborated with product and business stakeholders to define technical solutions aligned with operational requirements;
• Helped improve reliability, maintainability, and delivery speed across multiple engineering initiatives.

A real-time transport tracking platform for monitoring client transports with GPS-based location updates, status visibility, and incident reporting. The solution included a web application and mobile apps for custodians, giving clients real-time access to transport progress, checkpoints, photos, and analytics.

• Developed a real-time tracking application for monitoring transport location, speed, and status;
• Integrated GPS technology to provide accurate positioning throughout the transport journey;
• Built a web application that allowed clients to track multiple transports and generate reports and analytics;
• Developed a React Native mobile app for custodians to submit checkpoints, incident reports, and photos in real time;
• Integrated AWS services for data storage, real-time synchronization, streaming, and analytics;
• Built backend services with Node.js and serverless architecture patterns;
• Used AWS Amplify to streamline backend management, authentication, and authorization;
• Containerized and deployed backend services with Docker and AWS Fargate;
• Supported the solution with cloud infrastructure designed for scalability, reliability, and security.

A high-traffic digital platform with full-stack web applications, real-time dashboards, content management systems, and microservices-based solutions. The work focused on building maintainable frontend and backend systems while leading a small engineering team and improving delivery practices across multiple software projects.

• Led and contributed as both an Engineering Manager and individual contributor across the project;
• Managed the team while staying hands-on in development work;
• Built frontend components and pages for communication flow creation in the communication platform as a service;
• Worked with React, Java, and AWS to deliver and maintain the solution;
• Collaborated with the team to integrate the platform with different providers and ensure seamless functionality.