Front-End
I wrote interfaces: an industrial analytics portal in React + TypeScript. Even then, data visualized for decisions.
Open to new opportunities
AI Engineer — Buenos Aires, Argentina
FranciscoFilosi
I take AI agents and decision engines from prototype to daily operation.
From writing every step a bot takes to designing how an agent reasons: I automate work end to end, and I measure it.
20+AI agents in production
25+automation robots in production
57,000+products analyzed to validate a statistical model
700+hours of manual work freed per year
- Buenos Aires, Argentina
- 5+ years of experience
- Native Spanish · English B2
- Computer Engineer · MSc in AI in progress
- On-site, hybrid or remote
Decisions my systems make every day
Projects
The problem, what I did, what came out
AI agents
Enterprise AI Agent Platform
Blue Star Group's internal AI portal: one shared platform that solves once what every agent needs —reaching the data, searching, measuring and governing— instead of rebuilding it in each project.
20+ agents in production11 countries3 people led
PythonStreamlitDockerMCPQdrantSupabase+4
2025 — present
AI agents
Intelligent Markdown Engine
Decides which products to discount and by how much, leaving the decision ready to load into the company's ERP.
x5 feature growth in 6 months11 countries covered
PythonStreamlitSQL ServerLangChain+4
2025 — 2026
ML & Statistics
Seasonal Buy Optimizer
Decides how many units to buy of a product purchased only once per season: neither too few nor too many.
9 statistical steps2 inference methods by sample size
PythonStreamlitSciPySQL Server+5
Jun — Jul 2026
ML & Statistics
FOB Cost Predictor
Estimates what a product that does not exist yet will cost, so it can be budgeted and negotiated with suppliers before it is made.
57.000+ products analyzedC-level presentation audience
PythonStreamlitSQL ServerSciPy+4
Jun 2026
ML & Statistics
Cash-Audit Fraud Detection
Detects shortfalls and unusual behavior in the cash-ups of hundreds of stores, combining three machine-learning approaches.
3 ML approaches combined2 countries in production
Pythonscikit-learnXGBoostStreamlit+3
2025
AI agents
Job Hunter
A job-search copilot that scores every opening against my own professional knowledge base and drafts the application.
4 AI actions over the corpus11 API endpoints
Next.js 16React 19TypeScriptTailwind CSS v4+5
Jul — Aug 2026
The evolution
Five years moving up the abstraction layer
About
Taking AI from prototype to operation
I don't just train models: I understand the process, build the software, wire the data, talk to users and measure the outcome. I started in front-end, moved through RPA, BI and machine learning, and today I design agents and decision engines that run in production.
Production over demos
A system matters when someone uses it to decide. Everything I build aims at production, with evals, telemetry and support.
End to end
From raw data to the decision: process, data, model, interface and adoption. I don't stay in a single layer.
Numbers, not adjectives
Every project gets measured: SKUs validated, hours freed, agents running. And when a result has no number, I say so.
The stack
What I build with
Keywords
- Python
- SQL
- LLMs · Agentic AI
- RAG
- MCP
- LangChain
- OpenAI · Gemini · Claude
- CLIP
- Qdrant · Pinecone
- scikit-learn
- XGBoost
- Tableau
- SQL Server
- Supabase
- Docker
- FastAPI
- Streamlit
- Spring Boot
- Angular · React
- Automation Anywhere
What I work with
AI & Agents
RAG with citations in production · MCP (Python SDK · data-access pilot) · LLM-as-judge evals · GPT-4o vision · multimodal CLIP · LangChain · Claude API
ML & Statistics
scikit-learn and XGBoost in production · inference (Student's t, percentiles, Cohen's d) · clustering and UMAP · demand forecasting · anomaly detection
Data & BI
Heavy SQL Server · OLAP/MDX cubes · Tableau (~80 dashboards operated) · Qdrant, Pinecone, FAISS · Supabase/Postgres · EasyMorph
Backend & Platform
Python as main language · FastAPI with security by design · Java/Spring Boot (323 classes in TeamSync) · Docker on every deploy · pure-engine testing
Automation
Automation Anywhere certified (2022, renewed 2025) · 25+ robots · IDP/IQ Bot · C#/.NET metabots · own reusable framework
Technical leadership
Founded and led an AI Lab (team of up to 3) · org-wide standards and governance · trained the C-level in AI fluency · mentored juniors
Experience
Five years, three disciplines, two promotions
Jul 2026 — presentCurrent
AI Engineer · Blue Star Group
Promoted from Lab Lead to the IC track
- Design and ship LLM agents and statistical decision engines for retail —pricing, purchasing, assortment— across 11 countries.
- Own the platform layer the whole agent ecosystem runs on: MCP data access, evals, cost telemetry and security standards.
- The promotion moved me to the individual-contributor track: I stepped out of formal people management to work hands-on on the platform and its architecture.
PythonLLMs · Agentic AIMCPSQL ServerDocker
Apr 2025 — Jun 2026
Machine Learning & AI Lab Lead · Blue Star Group
- Founded the AI Lab: hired and led a team of up to 3 and built the internal portal that grew to 20+ AI agents in production.
- Shipped the flagship decision engines (pricing, purchasing, year-over-year analytics), integrated with the corporate ERP.
- Built an FOB cost predictor tested against 57,000+ SKUs and presented to C-level, plus ML fraud detection for cash audits.
- Ran corporate BI: ~80 Tableau dashboards for 11 countries, doubling automation and freeing 700+ hours a year.
Pythonscikit-learnLangChainTableauQdrantDocker
Jan 2024 — Apr 2025
Semi-Senior RPA Developer · WF Consulting
Promoted from Junior
- Led end-to-end automation projects for major LatAm companies, from discovery to production support, mentoring juniors.
- Set up the first full document-AI (IDP) environment for a leading travel company and led the Automation Anywhere v11 migration.
- Contributed to a 25+ production-robot portfolio — ~6,800 certified effective hours in total.
Automation AnywhereIDP · IQ BotC#/.NETPower BI
Jan 2022 — Dec 2023
Junior RPA Developer · WF Consulting
- Built 7+ production finance and ERP bots (JD Edwards) for the country's largest pharmacy chain.
- Co-created a reusable C#/.NET metabot framework that cut the team's development time.
Automation AnywhereC#/.NETJD Edwards
Jun — Dec 2021
Front-End Developer · Ampexa · 4i Platform
- Built the portal of an industrial analytics (OEE) platform in React 17 + TypeScript with i18n and role-based permissions, plus a C# Web API and Zebra scanner integration.
React 17TypeScriptC# Web API
Education
From math olympiads to a master's in AI
Master's in Artificial Intelligence
Universidad de San Andrés · 2026 — 2027 · in progress
Numerical optimization, statistical inference, stochastic modeling, algorithms and machine learning — while working full-time.
Computer Engineering
Universidad del CEMA · 2021 — 2025 · graduated
75–90% merit scholarship. Intermediate Systems Analyst degree (2024). Thesis: TeamSync.
Computer Science Technician
Instituto Santa Ana · Technical high school
8 years in national Math Olympiad finals; provincial runner-up and top-3 rankings.
Certifications
- Anthropic Academy11 courses on Claude, agents and MCP in 6 weeks (2026) — 8 with credential IDs
- IBM Data Engineering Professional CertificateCoursera · 2024 · 5 certified modules
- Automation Anywhere Certified Advanced RPA Professional2022 · renewed 2025
- Cambridge English: First (FCE)English B2 · 2023
Contact
Let's talk
I'm in Buenos Aires, Argentina, open to AI roles with an architecture component. If my profile fits what you're looking for, drop me a line.
franciscofilosi3@gmail.com