Senior TypeScript / Python AI Engineer

AI & Automated CRM

About the Project

Cloud-native autonomous agent platform unifying CRM and lead operations into a single event backbone. The agent layer handles scoring, intent parsing, and routing while requiring operator review before state changes or outbound actions execute.

• Sales

Responsibilities

– Designed a multi-step agent workflow across intake, enrichment, scoring, and routing decision nodes with defined input/output schemas, retry logic, and dead-letter fallback for unresolvable states; – Built an approval queue layer in Node.js where the agent halts after routing decisions and writes structured recommendations to SQS for operator review; – Built Pydantic schema enforcement so structured and auditable output is produced at every node before reaching downstream systems; – Implemented an immutable audit log recording agent recommendations, confidence scores, operator identity, decision timestamps, and resulting actions; – Designed escalation logic for pending approvals to reroute unreviewed items to the next operato;

Veterinary Health Platform

About the Project

High-stakes consumer health application requiring structured confidence scoring and policy-defined boundaries on agent output before recommendations reach end users.

• HealthTech

Responsibilities

– Designed a multi-agent pipeline in LangGraph and Temporal covering symptom intake, retrieval, differential reasoning, and recommendation generation; – Implemented persistent state between workflow nodes and explicit interrupt points when confidence dropped below threshold; – Built a NestJS orchestration layer exposing agent workflow endpoints, webhook consumers, and approval callbacks; – Built a document ingestion and retrieval pipeline using LlamaIndex with Pinecone, including chunking strategies and metadata filters; – Built a production recommendation engine returning structured verdicts, confidence scores, and written rationales with Pydantic enforcement; – Architected a FastAPI and NestJS backend with PostgreSQL and MongoDB data layer;

MarineIQ

About the Project

End-to-end service management platform covering the full vessel lifecycle with an AI-powered inspection and recommendation layer.

• Management

Responsibilities

– Built an AI-powered digital inspection engine converting unstructured technician field notes into structured inspection reports; – Generated categorized findings, recommended follow-up services, and customer-facing summaries with Pydantic validation; – Built a service recommendation layer analyzing vessel history, service intervals, and inspection findings for proactive maintenance recommendations ; – Designed a real-time scheduling engine with drag-and-drop job management and live field status updates; – Built customer communication automation including AI-generated status updates, digital inspection sharing, and e-contract approval workflows; – Delivered an AWS serverless backend using Lambda, API Gateway, RDS PostgreSQL,...

Cloud-Based Legal Practice Management & Billing Platform

About the Project

Law firm billing platform with Stripe payment integration, client invoice approval workflows, and real-time budget tracking.

• LegalTech

Responsibilities

– Built an invoice lifecycle from billing entry and client review to approval confirmation and Stripe charge execution; – Ensured Stripe charge API calls were made only after client approval events were recorded in the database; – Integrated third-party legal CRMs including Clio, RocketMatter, Action Step, and Practice Panther via API for data import; – Implemented token-based authorization and Redis-backed session management; – Implemented SNS/SQS notification infrastructure; – Contributed to database architecture design and full-stack feature deliver;

Vehicle Auction Platform

About the Project

Real-time auction platform handling transparent price discovery and secure transaction settlement across concurrent bidding sessions.

• Automotive

Responsibilities

– Built a WebSocket-based real-time bidding engine with live countdown timers, instant bid updates, and push notifications; – Integrated a third-party payment provider with escrow settlement infrastructure for transaction integrity across vehicle trades; – Built a fraud prevention layer with validation logic and data integrity checks at bidding and transaction level; – Designed a VIN Explorer Module for per-vehicle lookup, auction event binding, and historical transaction reference; – Deployed scalable AWS infrastructure with horizontal scaling and load balancing; – Implemented real-time observability for auction platform operations;

Automation Agency

About the Project

End-to-end LLM solutions delivery across multi-agent workflows, RAG pipelines, and function calling using Azure and open-source AI stacks.

• Management

Responsibilities

– Designed and shipped multi-agent systems using LangChain and CrewAI with structured output enforcement; – Built RAG pipelines using LlamaIndex and Pinecone for document indexing, retrieval, and context assembly; – Built a content moderation pipeline for a consumer UGC platform with category detection, severity scoring, and action routing; – Routed low-confidence moderation outputs to a human review queue instead of automated action; – Monitored false positive rate per moderation category on a weekly basis; – Integrated external APIs and internal developer tools into production AI automation pipelines;

Optical & Photonics Circuit Design Platform

About the Project

Engineering and scientific software platform for optical and photonics circuit design with AI-assisted development features and cloud collaboration.

• Scientific Research

Responsibilities

– Built an AI features layer for automated schema documentation generation, intelligent deduplication, and predictive component autocomplete; – Designed an AWS serverless collaboration backend with role-based project sharing and access metadata; – Integrated GDSfactory for automated routing and improved prototyping workflows; – Delivered a 60% faster prototyping cycle through GDSfactory integration and automation;

AI Assistant Platform

About the Project

Generative AI assistant platform requiring scalable cloud infrastructure, high availability, and fast deployment across frontend, backend, and plugin modules.

• Business Intelligence

Responsibilities

– Designed and maintained scalable cloud infrastructure for a generative AI assistant platform; – Migrated synchronous user-facing endpoints from Lambda to Cloudflare Workers; – Used Upstash Redis for session state and short-lived queue entries to reduce cold start latency on critical path requests; – Built CI/CD pipelines for frontend, backend, and plugin modules; – Required Playwright E2E, lint, typecheck, and unit tests to pass before merge and deployment; – Implemented security, access control, and compliance measures across multiple platforms; – Migrated infrastructure from CloudFormation to Terraform;