Avancini

Senior AI Engineer

Connect

[cv: cv.pdf]
[fingerprint: 0373C0DD571E7C6F12AA1A5E83B92CD78546F342]
[github: @avancini23]
[linkedin: Avancini Lara]

Profile Summary

Senior AI and software engineer focused on production LLM systems, AI platforms, and backend engineering. Experienced in taking AI applications from architecture through production, including model evaluation and optimization, APIs, workflow orchestration, cloud infrastructure, and developer tooling.

Experience

Senior AI Engineer @ TELUS Digital

Jul 2026 — Present
  • Design and build an LLM-assisted platform for identifying, evaluating, and ranking high-quality customer-support interactions across datasets containing hundreds of thousands of transcripts per month, including an automated feedback loop that validates LLM-generated SQL against user requirements and supports iterative corrections through natural-language feedback.
  • Designed a Next.js interface and FastAPI backend for reviewing, explaining, editing, and executing generated queries, replacing an initially model-specific prototype with an extensible, OpenAI-compatible API.
  • Expanded the platform with eligibility checks, quality scoring, and clustering workflows to further evaluate, filter, and organize candidate conversations.
  • Reduced LLM inference costs by more than 90% and made per-transcript processing approximately 3x faster by benchmarking models across accuracy, latency, and cost and moving workloads to more efficient domain-appropriate models.
  • Deployed and currently operate the application on Google Cloud Platform using Cloud Run, Cloud Build, Secret Manager, and OAuth 2.0, with ownership spanning frontend, API, LLM workflows, and production deployment.

Senior Software Engineer @ Telescope Partners

Jul 2025 — Jun 2026
  • Owned and evolved production workflows built with Apache Airflow, improving reliability, observability, and day-to-day operation of data and automation pipelines.
  • Designed and implemented LLM-powered extraction and classification pipelines using structured outputs, prompt engineering, validation, and failure handling for real-world inputs.
  • Improved LLM cost control through prompt consolidation, caching, token limits, and safeguards against runaway consumption.
  • Migrated data workloads from Tinybird-style implementations to PostgreSQL-backed architectures, focusing on query performance, maintainability, and operational simplicity.
  • Built reusable Python libraries and internal tooling that standardized workflow patterns and reduced duplicated implementation work across projects.

Senior Software Engineer @ Thoughtful AI

Jul 2024 — Jul 2025
  • Designed reusable internal Python libraries used across multiple projects, reducing duplicated implementation work and standardizing common development patterns.
  • Designed architecture for automation and backend systems with an emphasis on maintainability, performance, and production reliability.
  • Built and maintained healthcare automation products integrating internal systems, third-party applications, and external APIs.

Software Engineer @ Thoughtful AI

Jun 2023 — Jul 2024
  • Developed Python automation for healthcare workflows using RPAFramework, Selenium, browser automation, and reverse-engineered APIs.
  • Integrated healthcare systems without conventional APIs by analyzing application network traffic and reproducing required requests programmatically.
  • Automated workflows involving patient, financial, and operational data while maintaining data accuracy and handling production failure cases.
  • Monitored and improved deployed automations based on runtime behavior, errors, performance, and changes in external systems, collaborating with engineering and operations teams to diagnose issues.

Python Developer @ KarHub

2022 — Jun 2023
  • Developed ETL pipelines for collecting, transforming, and processing business data using Python and Pandas.
  • Built web scrapers, data collection systems, and RPA workflows using Selenium and other Python tooling.
  • Developed REST APIs with Flask and FastAPI and used RabbitMQ for asynchronous processing and service communication.
  • Built serverless workloads with AWS Lambda and used Apache Airflow for scheduling, orchestration, and monitoring of data pipelines.

Nonlinear Dynamics Researcher @ UNESP — Nonlinear Dynamics Laboratory

2017 — 2020
  • Developed computational models for nonlinear dynamical systems and translated an existing FORTRAN model to Python.
  • Performed numerical and statistical analysis of chaotic systems, including the Dissipative Standard Map.
  • Created scientific visualizations with Matplotlib and Gnuplot for analysis and presentation of research results.

Technical Skills

Programming

PythonSQLTypescriptAny, if it is needed.

AI & LLM Engineering

LLM Application DevelopmentLLM Evaluation & BenchmarkingStructured OutputsInformation ExtractionClassificationPrompt EngineeringModel SelectionFeedback LoopsInference OptimizationOpenAI-compatible APIs

AI Platforms & Models

Anthropic ClaudeOpen-weight LLMsSelf-hosted LLMs

Backend & Frontend

FastAPIFlaskREST APIsNext.jsReactPostgreSQL

Data & Automation

Apache AirflowPandasETLRPAFrameworkSeleniumRabbitMQWeb Scraping

Cloud & Infrastructure

Google Cloud PlatformCloud RunCloud BuildSecret ManagerOAuth 2.0AWS LambdaAmazon S3Docker

Education & Certifications

Bachelor's Degree in Physics

2020 — 2022

Federal University of Rio Grande do Sul (UFRGS)

Physics — transferred before completion

2017 — 2020

São Paulo State University (UNESP)