Data Platform

• Created Wellfound’s data platform, powering company-wide analytics and in-product data features.

• Operate large-scale data pipelines in Dagster and dbt, ingesting and modeling multi-source datasets into AWS Redshift and an AWS-based Data Lake.

• Expose and evolve real-time data APIs using FastAPI and Pydantic validation

• Maintain and improve MLOps infrastructure: LLM Proxy layer with real-time logging, uptime, performance, and cost observability; integrated Braintrust for LLM evaluation.

• Scale and optimize HuggingFace ML classification and embedding pipelines, serving billions of predictions in production.

Candidate Sourcing Service

• Aggregate candidate and company data from diverse internal and external sources.

• Transform, normalize, and ML-enrich data (classification, tagging, semantic similarity).

• Index and serve 1B+ candidate profiles and millions of companies in Elasticsearch for high-performance, recruiter-facing search and matching.

Tech: AWS (S3, Redshift, DMS, Athena, Lambda, CloudWatch), Elasticsearch, Python, FastAPI, Pydantic, HuggingFace, dbt, Dagster, Datadog, SQL, LLM

• Built and led Envoy’s Data Engineering function from the ground up, enabling self-serve analytics, machine learning, and experimentation across the company.

• Managed and expanded a cross-functional data team to 9; introduced eng best practices (QA, testing, observability) to improve reliability and trust in analytics.

• Designed and implemented customer cohort feature adoption models in dbt + Looker, delivering actionable product insights.

• Developed a scalable event tracking ingestion pipeline (Segment → AWS Glue Data Lake → Redshift) with downstream data models powering analytics, A/B tests, and reporting.

• Delivered embedded, customer-facing analytics inside the Envoy product using AWS DMS, Django, Pandas, React, and Vega.

• Integrated Salesforce and Intercom with internal analytics pipelines, syncing millions of user attributes for sales and support.

Tech: AWS Glue, Redshift, Data Lake, dbt, Python, Django, React, Datadog, Segment, Amplitude, Looker

After 10 years as a leader at 8tracks, I moved to an advisory position.

I contributed to data and ML initiatives until 8tracks was acquired, then led the technical transition to the new owners.

• Co-founded 8tracks, scaling it into a beloved music discovery platform with 8M monthly active users and a vibrant, loyal community.

• Built and led a passionate, high-performing EPD team from 1 → 12, fostering creativity, collaboration, and a deep connection to our mission.

• Engineered a high-performance service, consistently maintaining sub-100ms API responses at scale.

• Pioneered data science-driven music recommendations using word2vec within months of its publication.

• Developed a highly scalable in-house product analytics and experimentation platform, processing 1B+ events/month to drive product decisions.

Tech: Ruby, Redis, Redshift, StatsD, D3.js, SQL, AWS, sklearn

• Established real-time event observability with StatsD → Graphite/Grafana to monitor deliveries, inventory, and driver location.

• Integrated Domino Data Lab to enable data scientists to independently develop and deploy ML models.

• Consolidated data warehousing into Redshift and standardized LookML for consistent, reliable analytics.

• Built evaluation and visualization tools for monitoring delivery ETA prediction model performance.

Tech: StatsD, Graphite, Grafana, Redshift, Looker, Python, Domino Data Lab