jason blahovec
about · personal

About me.

I'm a GCP-native ML systems engineer. I build production ML platforms end-to-end on Google Cloud — Cloud Run inference, Vertex AI training, BigQuery warehouse, Firestore state, Cloud Build + Artifact Registry on the deploy path.

PARROT is the working proof — a multi-tenant agentic platform with a model registry, MCP tool catalog, persona-driven chat, and an OWASP-tested adversarial eval harness. Caveats are content. Every accuracy number wears its sample size and its bias label. No "AI-powered." No "revolutionary." Just the number and what's hiding inside it.

work history
5 entries · most recent first
periodrolesummarystack
2025 — present
Sr Data Scientist · AI Research & Development
NRG Energy · Remote
Selected to represent the Data Science organization on NRG's Enterprise Data & AI R&D team after the late-2024 reorganization. Designed a LangGraph multi-agent that generates PowerPoint slide decks; championed LangFuse adoption for LLM tracing across the team; piloted Argilla as a UI for dataset annotation; contributed to the AWS → GCP migration of CLV, Cash Flow, and Call Volume Forecast models.LangGraphLangFuseArgillaGCP
2022 — 2025
Sr Data Scientist · Enterprise Data & Analytics
NRG Energy · Remote
Promoted to Senior Data Scientist after delivering NRG's Customer Lifetime Value model for business customers. Designed and deployed a Generative AI/NLP pipeline to analyze employee sentiment across HR interviews; built five supervised call-volume forecasting models for customer service desks; developed a predictive model for cash receipts; engineered a regional power-outage analytics system.GenAINLPForecastingPredictive modeling
2021 — 2022
Data Scientist
NRG Energy · Pittsburgh, PA
Joined the newly formed Enterprise Data and Analytics team after Direct Energy's January 2021 acquisition by NRG. Led development of the B2B Customer Lifetime Value model, integrating data to provide strategic insights and support financial planning.CLVB2B analyticsPredictive modeling
2018 — 2020
Data Engineer
Direct Energy Business · Pittsburgh, PA
Developed and enhanced the company's AWS-based data lake within the B2B marketing team. Integrated the data lake with Google AdWords, Invoca, and Bing Ads; inherited and refined the CLV Business ETL by migrating it to AWS, preparing inputs for predictive models I later implemented as a Data Scientist; conducted weekly cloud-computing training sessions for junior developers.AWSETLMarketing APIsMentorship
2016 — 2018
Sr Data Analyst
Direct Energy Business · Pittsburgh, PA
Developed and implemented a Salesforce-integrated framework to automate the distribution of Qualtrics surveys to customers post-interaction, improving feedback collection and analysis across customer service and renewal processes.SQLSalesforceQualtricsCustomer feedback
Earlier (2011 — 2016): Senior SQL Developer and Senior Pricing Analyst at Thermo Fisher Scientific; e-Commerce Reporting Analyst at DICK's Sporting Goods (Order Master data model); Demand Planning Analyst at Thermo Fisher Scientific.
skills · grouped
currently shipping · parrot
Cloud RunVertex AIBigQueryFirestoreMCPNextAuthOWASP LLM eval
top skills
LLMOpsArtificial IntelligenceCloud Computing
agentic · llm
LangGraphLangFuseLangChainGeminiArgillaGenerative AINLP
cloud · data
Google CloudAWSApache SparkPySparkDatabricksBigQuery
ml · modeling
Supervised learningUnsupervised learningFeature engineeringHyperparameter optimizationForecastingCustomer Lifetime Value
languages · tooling
PythonPySparkSQLETL pipelines
open source
4 published · BigQuery ingestion libraries
statcast-bigqueryv0.4.0
Idempotent Statcast → BigQuery ingestion, with first-class docs for SQL/LLM agents and round-trip validation against Baseball Savant.
118 tests · Savant round-trip verification · resumable chunked backfill
PythonBigQuerypybaseballCLI
yfinance-bigqueryv0.1.0
Idempotent Yahoo Finance OHLCV → BigQuery ingestion across 5 intervals, with first-class docs for SQL/LLM agents and internal-consistency verification.
126 tests · 5 intervals · BigQuery-internal verification
PythonBigQueryyfinanceCLI
nhl-bigqueryv0.2.2
Idempotent NHL play-by-play (with on-ice arrays merged from shift-charts) → BigQuery ingestion, with first-class docs for SQL/LLM agents and verification against the NHL public API.
126 tests · 6 tables in lockstep · on-ice arrays from shift-charts
PythonBigQueryNHL APICLI
nhl-hut-bigqueryv0.1.1
Idempotent EA NHL Ultimate Team ratings → BigQuery snapshots (skater + goalie endpoints unified), with a cross-source resolver that links HUT cards to NHL player IDs.
56 tests · skater + goalie snapshots · normalized-name card↔player resolver
PythonBigQueryEA NHL HUTCLI
education
Georgia Institute of Technology
M.S. Computer Science — Machine Learning
Carnegie Mellon University
B.S. Mathematics
currently shipping
currently shipping
PARROT
An agentic MLB analytics app — Statcast, umpires, games & odds through a model registry, MCP tool catalog, persona-driven chat, and an OWASP-tested adversarial eval harness. End-to-end on Google Cloud.
MLB analyticsStatcast + odds6 agents40 MCP toolsGCP-native
Enter the app →