Skip to content

Résumé

The short version

Roles, results and skills on one screen. The PDF version is formatted for applicant tracking systems.

Summary

Lead AI Engineer at BlackPrint Technologies, building LLM agents, GeoAI platforms and production ML. At BlackPrint: mireporte.ai market studies across 11 metro areas, GeoAI over 2.5M+ urban blocks and payment-acceptance scoring for ~5.5M merchants. For the Law Offices of Manuel Solis: legal LLM agents that cut document turnaround ~8×. Medical AI research, UTSA. BA Data Science, Berkeley; Master's in Applied AI, Tec.

Experience

  1. 2025 – Present

    Current

    Lead AI Engineer

    BlackPrint Technologies · Remote

    • Led the map copilot's redesign into a tool-grounded agent (Copilot v2): 38 FastMCP tools with per-tool authorization, 53 user-confirmed UI actions and a 778-question blind-annotated gold eval set.
    • Architected a production WhatsApp agent that qualifies real-estate leads and books appointments via OpenAI tool calling, with a deterministic slot validator, human takeover and 13 golden evals.
    • Built GromAI Report Studio, grounding Claude site-report narratives behind a numeric firewall, claim ledger and anti-leak guard with deterministic fallback; 15 eval suites and 2,809 tests.
    • Built an expansion-planning OS demoed to three chains: Claude tool calling over deterministic services (87% prompt-cache hits, 58/58 Q&A audit) and a finance engine within +0.27% of public disclosures.
    • Led mireporte.ai, which turns a map point into a 25-section market study, from 3 to 11 metro areas: per-city data contracts that fail the build on drift, sha256 manifests, 19 ADRs and 5,043 tests.
    • Built an LLM-powered GeoAI platform for 2.5M+ urban blocks and a multi-brand site selector (63,724 blocks, 117 municipalities) whose v1.2 sales model cut leave-one-out MAPE from 106% to 22%.
    • Scored the payment-acceptance propensity of ~5.5M merchants for a global payments network (45 GB Postgres to a 70 MB Parquet via DuckDB) and reconciled 230,553 business records with MiniLM embeddings.
    • Built logistics optimizers: an exact MILP (HiGHS) partner selector on a 4,851-pair OSRM drive-time matrix for a digital pharmacy, and VRP routing for a pharma distributor's 3,100+ pharmacy pilot.
  2. 2025 – Jul 2026

    Lead AI/ML Engineer (Consulting)

    Law Offices of Manuel Solis · Remote

    • Led custom LLM agents for T-Visa declarations and cover letters, cutting turnaround from ~2 hours to ~15 minutes per document (~8× faster).
    • Built a meeting-intelligence platform with Speechmatics transcription, GPT-5 structured summaries, Pinecone search and group-based RBAC.
    • Automated email-to-case database ingestion and assignment.
  3. 2025 – Jan 2026

    Visiting Student Researcher (Medical AI, volunteer)

    UT San Antonio · San Antonio, TX

    • Developed a multi-agent medical QA system grounded in PrimeKG (4M+ nodes, 33M+ edges) and Neo4j.
    • Reached 54.69% QA accuracy (+24.24 pp) with multi-hop reasoning and systematic evaluation.
  4. 2023 – 2025

    AI/ML Engineer

    STRTGY · Remote

    • Built a RAG assistant for industrial real estate (Claude 3.5 Sonnet, FAISS; +90% response relevance, 85% faster queries) and GPT-4 Vision competitor mapping in ArcGIS (+82% decision confidence).
    • Built adaptive demand forecasting for a beverage-industry client (Prophet or SARIMA chosen per SKU history, with inventory optimization): +42% forecast accuracy and 30% lower inventory cost.
  5. 2024 – 2025

    AI Engineer (Contract)

    PRODENSA · Remote

    • Built OCR + Llama 3.3 extraction for invoices, packing lists and customs documents (95% accuracy, 98% less processing time) with FastAPI, React/Vite and async PDF/CSV/XLSX ingestion.
  6. 2022 – 2023

    Data Scientist

    BlackPrint Technologies · Berkeley, CA

    • Delivered property analytics with Google Maps geocoding (92% accuracy) and pandas/SQL ETL pipelines with 99.9% deliverable accuracy.

Selected projects

See all projects

Skills

GenAI / LLM
RAGAgentsLangGraphLangChainFine-tuning (LoRA)EmbeddingsFAISSPineconeSpeechmatics
LLM platforms
OpenAI (GPT-4o / GPT-5)ClaudeGeminiLlama
ML / DL
PyTorchTensorFlowscikit-learnXGBoostLSTMForecasting
Geo
GeoPandasMapLibre GLMapbox GLTurf.jsH3ArcGISQGIS
Data
PythonSQLBigQueryRedshiftpandasNumPyNeo4jMongoDB
Backend / Cloud
FastAPINode.jsTypeScriptNext.jsGoogle CloudTerraformMLflowDocker

Certifications

  • Enterprise Data Science in Practice

    IBM SkillsBuild · Mar 2025

    Verify
  • Data Analysis with Python

    Cognitive Class (IBM) · Mar 2025

    Verify
  • Sutardja Center for Entrepreneurship Certificate

    UC Berkeley, Sutardja Center for Entrepreneurship & Technology · Fall 2022

PDF preview