Eva Agapaki, Ph.D.
Machine Learning Engineer and Research Scientist (10 years experience), expert in building production-grade 0-1 ML products specializing in LLM-based agents (tool use, context management), Agentic Workflows, and retrieval (RAG).
Current Experience
Senior Data Scientist
IKEA (Contract) • New York, NY
Nov 2024 - Present
- Scope, research and develop agentic and data science projects for product and engineering to improve recommender systems’ retrieval and relevance metrics.
- Developed an agentic flow for automated data labeling using the Agent Development Kit (relationship inference based on unstructured data) with 70% expected reduction in manual labeling focusing heavily on tool use, prompt engineering, and guardrails for reliable outputs.
- Deployed semantic product search across 8 global markets and an AI Judge evaluation framework that accelerated offline experimentation and drove an increase in Average Order Value.
- Built a harness optimization loop to validate semantic deployments and fine-tune open-weights SLMs (e.g., Qwen) for optimal quality, latency, and cost.
- Design and implement A/B testing experiments for improving search retrieval (SIGIR 2026 paper accepted for spotlight oral presentation on hard negative mining methods for contrastive learning).
- Built a production-scale knowledge graph integrating 10M+ structured and unstructured data (images, PDFs, 3D assets) using Neo4j and vector DBs to improve retrieval of product recommendations (led end-to-end development from concept to POC to production-scale)
- Productionized ML prototypes by partnering with engineering and MLOps to integrate PyTorch/Vertex AI models into IKEA’s personalization ML stack.
Past Experience
Founder & Chief Scientist
Hatch Labs • New York, NY
Sept 2023 - Nov 2024
- Developed and deployed a portfolio of RAG-based LLM agents and conversational chatbots for pre-seed and seed stage startups focusing on high-accuracy retrieval and cost-optimized inference (Sample portfolio).
- Developed LLM evaluation frameworks to derisk product launches. Consulted on AI Governance, ensuring model compliance with emerging regulations.
Head of Product & Data Science
Next Millennium Media • New York, NY
Jan 2023 - Sept 2023
- Led a 20-person engineering and data science team (15 engineers, 2 DS, 2 PMs, 1 UX) building sales-side auctions, A/B testing, and SSP/DSP product launches.
- Led data-driven development and experimentation for consumer-facing ad-tech products, increasing user adoption by 80% and driving +150% revenue growth in Q2 2023 through analytics-based feature prioritization, and rapid iteration.
- Delivered ML-powered models for ad revenue optimization, audience targeting, and ad-quality scoring—resulting in $1.5M in incremental revenue through model-informed product features and algorithmic improvements.
Founding Scientist & AI Lead
Tenet Twin LTD • London, UK
Sep 2019 - Mar 2023
- Built and launched an AI-powered Digital Twin platform, engineering large-scale ingestion pipelines for real-time sensor data and computer vision video streams to deliver predictive operational insights for enterprise oil & gas environments.
- Scaled ML infrastructure from proof-of-concept to paid B2B licensing, validating model performance on live operational data and rapidly iterating with engineering.
Assistant Professor & Director of Digital Twins Lab
University of Florida •
Nov 2021 - Feb 2023
- Partnered with the U.S. DOT to build a multimodal evaluation system for infrastructure maintenance, engineering pipelines to fuse drone video/image feeds with unstructured text from engineering logs to automate structural state classification.
- Collaborated with DFW Airport to ingest, index, and process large-scale, real-time data from multiple physical sensors, building ML models to optimize terminal operations and active learning workflows to reduce passenger congestion.
Senior Software Engineer
PTC • Boston, MA
Jan 2020 - July 2021
- Led the end-to-end development of an ML-powered API and AR/VR computer vision product, integrating Azure Custom Vision with PTC's asset management platforms to deliver real-time object detection.
- Transformed an early-stage AR SaaS prototype into an award-winning, ML-driven product that boosted factory-floor training efficiency through real-time CV inference and interactive guidance—recognized as a “Best Innovation” product at PTC.
- Built and scaled a cross-functional ML product team from 0→32 (Data Scientists, Engineers, UX), driving fast experimentation, model evaluation, and MVP delivery aligned with executive leadership (link).
Senior Data Scientist
PTC • Boston, MA
Jan 2020 - July 2021
- Led the end-to-end development of an ML-powered API and AR/VR computer vision product, integrating Azure Custom Vision with PTC platforms to deliver real-time object detection for industrial training and operations.
- Built and scaled a cross-functional ML product team from 0→32 (DS, Engineers, UX), driving fast experimentation, model evaluation, and MVP delivery aligned with executives.