About
Hi, I’m Maxwell — a Principal Engineer at HTC DeepQ. I build machine learning systems at the intersection of natural language processing and computer vision. Recently, I’ve focused on LLM‑based products for data extraction, clinical chatbots, and medical‑report understanding. I also lead a team of engineers. Previously, I was a machine learning researcher in the Computational Imaging Lab at Sheba Medical Center (Newsweek top‑10 worldwide).
I hold an M.Sc. in Electrical & Electronic Engineering from Tel Aviv University, advised by Prof. Nahum Kiryati and Dr. Arnaldo Mayer. My work spans LLMs, outlier detection for CV/NLP, semi‑supervised learning via knowledge distillation, and flow‑based generative models.
Highlights
- 2nd place, 1st HTC Artificial Intelligence Hackathon (2023, 60+ business units competing)
- Spotlight presentations: MICCAI 2022, MIDL 2020
- [2018–2020] M.Sc., Tel Aviv University — GPA 90/100 (Accelerated Track)
- [2015–2019] B.Sc., Tel Aviv University — GPA 92/100 (Magna Cum Laude)
Timeline
-
Principal Engineer — HTC DeepQ
- Architected hybrid‑reasoning LLMs with reinforcement‑learning post‑training, achieving state‑of‑the‑art accuracy on complex clinical‑language tasks.
- Deployed a production pipeline that processes ~200,000 nursing reports each month—validated by nursing professionals—while outperforming GPT‑4o in accuracy, running 30× faster, and delivering 5,000× lower cost.
- Built organ‑specific summarizers for prostate and breast‑cancer cases, fusing long‑form, multi‑domain clinical data (lab, pathology, imaging, narrative notes) into concise, physician‑validated synopses.
- Lead a team of engineers and interns exploring data‑extraction and information‑validation techniques.
- Partner daily with full‑stack engineers, product managers, and physicians to deliver production‑grade AI solutions.
-
Senior Engineer — HTC DeepQ
- Developed custom LLM models for clients in healthcare, government, media, and legal sectors.
- In charge of Large Language Model (LLM) research & development (pre‑training, fine‑tuning, evaluation, and deployment).
- Owner of DeepQ AI Platform v2 (feature development and maintenance).
- Invented internal data cleaning tools for vision and language data management applications.
- Developed specialized CAD model for simultaneous lung nodule classification and localization (featured model in AI Platform v2.3).
- Authored 2 papers and 6 patents.
-
Machine Learning Researcher — Sheba Medical Center
- Led the development of a COVID‑19 lung CT screening system in Sheba [underwent pilot tests]
- Developed an automated knee MRI diagnosis algorithm suitable for triage (tested and verified by the Head Radiologist of the MSK Department)
- Implemented a text analytics algorithm for labeling breast MRI reports written in Hebrew and English
Publications & Patents
Exploring Health Misinformation Detection with Multi‑Agent Debate
WASP Workshop - IJCNLP-AACL, 2025
Data classification method for classifying inlier and outlier data
US Patent US12292919B2
Data classification method for filtering outlier text data
US20250077552A1 (patent pending)
Medical image detection system, training method and medical analyzation method
US20230316511A1 (patent pending)
Automated Knee Injury Detection and Localization using 3D MRI Images
Master’s Thesis - Tel Aviv University, 2020
