Jangho Seo
Jangho Seo

Hello, I'm

Jangho Seo

ML Engineer & Data Scientist

I build ML systems judged by what they change in real decisions — insurance claims, KPIs, course completion. First-author publication in bioinformatics, recommendation systems running across about 800 client sites, and now decision intelligence at DEIN.

Based in Seoul — happiest shipping from anywhere with a good view.

01. About

The part of ML that's easiest to skip is the last mile: making a model actually change a decision. That's the only version I find worth doing. Right now that's decision intelligence at DEIN — delivering decision solutions for finance and insurance, judged by business outcomes rather than accuracy on a held-out set.

I've worked across applied ML — recommendation and knowledge tracing for learning platforms, computer vision on real construction sites, and peptide-identification research I published as first author. The throughline isn't one technique; it's carrying a problem from "no one's framed this yet" to something that ships and holds up in the real world.

The domain keeps changing — bioinformatics, edtech, fintech — but the question doesn't: how does something that works in a notebook move something real in the world?

My current research interests lie in Human-Centered AI — recommender systems, user behavior simulation, causal inference, and trustworthy AI.

Trajectory

  1. 2026 Joined DEIN, spun off from Ailys, to focus on decision intelligence for finance and insurance.
  2. 2025 Published NovoRank as first author in J. Proteome Res.; moved from edtech into insurance and credit-risk ML at Ailys.
  3. 2024 Super Rookie Award at Tmax A&C, given to outstanding first-year employees across its affiliates.
  4. 2023 Joined TmaxRG to build learner diagnosis and course recommendation.
  5. 2022 Presented NovoRank at the 70th ASMS Conference in Minneapolis; finished the M.S. in Artificial Intelligence at Hanyang.
  6. 2020 Began the M.S. at Hanyang, joining Prof. Paek's bioinformatics lab.
  7. 2019 First ML research at Inha's Construction Systems Lab; co-authored a paper for KICEM.
  8. 2014 Started a B.S. in Civil Engineering at Inha, later adding a Software Convergence double major.

Not on the resume, but part of the picture — more about me →

02. Education

Hanyang University — M.S. in Artificial Intelligence

Sep 2020 – Aug 2022

Advised by Prof. Eunok Paek. GPA 3.91 / 4.0. Thesis: NovoRank — Machine Learning Based Post-processing for Performance Improvement in De Novo Peptide Sequencing.

Inha University — B.S. in Civil Engineering; B.S. in Software Convergence Engineering (interdisciplinary major)

Mar 2014 – Feb 2020

GPA 3.60 / 4.5 overall, 4.30 / 4.5 in CS coursework (data structures, algorithms, operating systems, networks, databases, data mining).

Honors & Activities

Skills & Languages

Programming
Python, Java, SQL
ML & Data
PyTorch, TensorFlow, scikit-learn, Polars, Airflow, MLflow
Backend & Cloud
Django (REST Framework), FastAPI, Docker, AWS
Languages
English (professional working proficiency), Korean (native)

03. Experience

Professional

DEIN — Product Manager

Aug 2026 – Present

Spun off from Ailys to focus on decision intelligence for financial and insurance clients. I plan platform features that simulate policy outcomes, deliver scoring models, and release and maintain the ML platform for enterprise clients in Korea, Japan, and Thailand, including a 94% cut in monthly server cost.

Details →

Ailys (now DEIN) — Machine Learning Data Scientist

Apr 2025 – Jul 2026

Led KPI Maker, an LLM agent that turns domain experts' plain-language KPI descriptions into verified code, producing correct values for all 12 test KPIs. Built a multi-label claim payout model for a life insurer, predicting which of 1,580 diseases and 14 benefit types will be paid within two years (93% recall, 88% specificity). Also built underwriting KPI data marts and researched reject inference for credit applicants without credit history.

Details →

TmaxRG — Researcher, Diagnosis and Recommendation Team

Jan 2023 – Nov 2024

Developed a course recommender for Samsung Multicampus using learning histories and BERT-based title similarity, raising enrollment conversion by 5% and completion by 25% over the previous recommender. Improved knowledge-tracing models with new features from WAPL Math, led a proficiency-based problem recommender for digital math textbooks, and built Airflow pipelines for retraining and batch inference serving about 800 client sites. Helped write and present a KEIT research proposal on peer learning for worker reskilling, funded with ₩2.7B in government grants over three years.

Details →

Research

Bioinformatics and Intelligent Systems Lab, Hanyang University — Graduate Research Assistant

Aug 2020 – Dec 2022

Led NovoRank, an open-source tool that re-ranks de novo peptide sequencing candidates with two-step spectral clustering (MS-Cluster, DBSCAN) and a Siamese CNN-BiLSTM model, improving precision by 4.6 and recall by 4.5 percentage points on average across three de novo sequencing tools. Published as first author in J. Proteome Res. (2025) and presented a poster at ASMS 2022. Also took part in a semi-supervised PSM rescoring project focused on the 1% FDR region.

The Construction Systems Laboratory, Inha University — Undergraduate Researcher

Jul 2019 – Feb 2020

Took part in a digital-twin-based safety rule checking project for indoor construction sites, funded by the National Research Foundation of Korea. Built an image dataset of small construction tools, compared augmentation methods (brightness change with zoom worked best), and developed an EfficientNet-B0 tool classifier reaching 94.8% test accuracy.

04. Publications

NovoRank: Refinement for De Novo Peptide Sequencing Based on Spectral Clustering and Deep Learning

Jangho Seo, Seunghyuk Choi, Eunok Paek — J. Proteome Res. 2025, 24, 2, 903–910

Read paper →

Image Augmentation for Small Object Detection on Indoor Construction Site

C. Jeon, Jangho Seo, E. Lee, M. Chung, K. Lee, D. Shin — KICEM Annual Conference, Goyang, Korea, Nov. 2019 (in Korean)

Talks & Presentations

Post-Processing Framework Using Deep Learning to Improve De Novo Peptide Sequencing Results

Jun 2022

Poster · 70th ASMS Conference on Mass Spectrometry and Allied Topics, Minneapolis, MN

05. Projects

Featured

Decision Intelligence

Decision Intelligence · Insurance & Finance

At DEIN (formerly Ailys) I build ML judged by business KPIs, not accuracy on a held-out set — a claim payout model for a life insurer and an LLM agent that turns KPI descriptions into verified code.

93%

Claim payout recall

88%

Specificity

12/12

KPI Maker test KPIs correct

NovoRank workflow

NovoRank

Bioinformatics · Deep Learning · Clustering

Conventional de novo peptide sequencing tools score spectra in isolation and frequently misidentify peptides. I built a two-step clustering and deep learning re-ranking pipeline that cross-checks candidates against similar spectra.

+4.6 pp

Avg. precision gain

+4.5 pp

Avg. recall gain

Across three state-of-the-art tools

More projects

Personalized Learning at Scale

Recommendation Systems · Knowledge Tracing

Course recommendations on a corporate learning platform were generic and didn't reflect learner progress. I built a recommendation model on sequential interaction data and learning histories and shipped it through an Airflow pipeline, lifting course enrollment (CVR) by 5% and content completion by 25%.

Image augmentation for construction-site objects

Construction Site Hazard Detection

Computer Vision · Image Augmentation

Recognizing small hazard objects on construction sites is data-hungry, but labeled images are scarce. I studied which image-augmentation strategies actually help, lifting a CNN's accuracy from 85.7% to 87.6% — and found rotation hurts more than it helps for these objects.