Software Engineer · Data Engineering and Applied AI
Felipe Duenas
I build software that enlivens data for intelligent decisions.
I'm a software engineer who specializes in data. At AWS that means the data infrastructure and pipeline logic for a supply chain automation platform, along with the agentic work so the platform's agent doesn't just sound better, it drives results.
Statistics & Data Science, UCLA '27 · SF Bay Area | Los Angeles
Work
What I've built
A few projects, each with the full story behind it: what broke, what I built, what I'd change.
Amazon Web Services June 2026 — Present Making 3,000 dashboards answerable by an AI agent
Airflow ETL from Tableau workbooks to an AI-ready knowledge base
- 3,000+
- Tableau workbooks in the pipeline
- 85%
- Increase in queryable data
- Python
- Airflow
- Tableau REST API
- Amazon S3
- +2
Amazon Web Services 2026 Dashboards from a sentence, in 3 minutes
Strands Agent + MCP tooling that builds dashboards from a prompt
- 4 days → 3 min
- Median dashboard development time
- 15
- Features scoped with stakeholders
- Python
- Strands Agent
- MCP
- Tableau
- +2
UCLA Trustworthy AI Lab Jan 2026 — May 2026 An agent that does the data science, not just the SQL
71% on BIRD, eight points above GPT-4o
- 71%
- Task completion on the BIRD benchmark
- +8 pts
- Above GPT-4o on the same benchmark
- Python
- AWS Lambda
- API Gateway
- AWS CDK
- +3
Amazon Web Services Jun 2025 — Sep 2025 Forecasting data center power: $400K in savings
Pipeline & ensemble model on 500M+ daily telemetry events
- 1.9%
- Forecast MAE, against a 6.3% baseline
- $400K
- Cost-saving initiative informed
- Python
- Redshift
- dbt
- SQL
- +3
Experience
Where I've worked
-
Amazon Web Services
Seattle, WA · Arlington, VA
2025 · 2026 —
-
UCLA Trustworthy AI Lab
Los Angeles, CA
2026
-
Acer
San Jose, CA
2024
-
June 2026 — Present
Now
Software Engineer Intern
Amazon Web Services · Data Platform · Seattle, WA
- The org's operational knowledge lived in 3,000+ Tableau dashboards that no software could read. I built the pipeline that turned them into something an AI agent can answer from on our internal supply chain automation platform serving 5,000+ users. 85% more of the org's data became reachable, and the accuracy gain traced to the data, not the prompt.
- Then went the other direction: tooling on LangGraph and MCP that takes a sentence and gives back a dashboard. Four days of work, down to three minutes.
-
Jan 2026 — May 2026
Backend & Data Engineer Intern
UCLA Trustworthy AI Lab · Agentic Data Science · Los Angeles, CA
- Built an agent that plans and runs a whole analysis from a plain-language question. It completed 71% of BIRD benchmark tasks, eight points above GPT-4o.
- The interesting part was the harness that told us why it failed: 34 schemas, 120+ question types, and three architectural fixes worth +11% that we'd never have found from a pass/fail score.
-
Jun 2025 — Sep 2025
Business Intelligence Engineer Intern
Amazon Web Services · Arlington, VA
- Planners at government sites were coordinating builds around power numbers padded by buffers on top of buffers. An XGBoost and Random Forest ensemble, forecasting from historical draw and peak events, cut that error from a 6.3% baseline to 1.9%: tight enough to recalculate the planned values and fit more racks into the same footprint, with 99.9999% uptime held.
- Built the ELT pipelines for 500M+ daily telemetry events through dbt and Redshift, then one planning dashboard in place of the seven sources my team was reading by hand, and automated S3 Glacier demand tracking off a manual spreadsheet. Weekly reporting went from six hours to thirty minutes.
-
Jun 2024 — Sep 2024
Data Analyst Intern
Acer · San Jose, CA
- 30+ years of records were scattered across siloed systems, so I created a single data model in PostgreSQL that helped increase data accessibility by 125% for Acer's legal team.
- I helped cut report preparation time for General Counsel by creating automated Tableau dashboards, and I elevated licensing discrepencies through analyzing trends from merger and financial docs.
Education
Expected March 2027
University of California, Los Angeles
B.S. Statistics & Data Science · Minor in Data Engineering
Lab
Code you can read
The work above is internal, so none of it is open. These are smaller things I built on my own time, and the repos are public.
-
Data eng · Data Science
Event Demand — Bayesian Inference
A real pipeline, end to end: streamed real-world event data into an S3 data lake, staged into Snowflake, and modeled in dbt. Derived signals and insights using Bayesian statistical methods and the R language, with the whole thing on an Airflow DAG.
- Airflow
- dbt
- Snowflake
- S3
- Docker
- R
-
Data Science · Applied AI
Autonomous Reasoning Agent
Developed a multi-agent system programmed to perceive, reason, act, and observe. Implemented ideas such as social-chain-of-thought and self-correction, and developed a self-evolution system through RL. Deployed in online tournament to play social and conversational games, receiving a top 3 finish.
- Python
- ReAct
- Claude
- MCP
- Agent Harness
-
Applied AI · full stack
Asset Edge
A financial research application to serve real-time stock information, news, and forecasted insights derived from an LSTM neural network I implemented. Architected a Generative AI assistant supported by a RAG system for SEC filings and news articles, FinBERT for sentiment analysis, and FastAPI behind a React front end.
- FastAPI
- React
- LangChain
- ChromaDB
- PyTorch
Toolkit
What I reach for
Things I've actually shipped with, not everything I've ever opened.
- Languages
- PythonSQLJavaRBash
- Cloud
- AWSGCPRedshiftSnowflakeS3LambdaBedrock
- Data platform
- AirflowdbtSparkDockerTerraformCDK
- Applied AI
- ClaudeLangGraphMCPRAGRetrieval evaluationVector databases
- Modeling
- scikit-learnPyTorchForecastingAnomaly detectionNLP
Contact
Don't hesitate to reach out. Students, professionals, or people who want to talk about shooting data centers into space.