D. Sery
Data Scientist

Daniel Sery

Analytics and machine learning for finance, energy, and healthcare.

MS Data Science, UVA · Open to relocation
01 / About

A short introduction.

I build end-to-end systems — from ingest and validation through modeling, evaluation, and the stakeholder-facing tools that make the work usable. I’m comfortable on both ends of the stack: writing SQL stored procedures one day, fine-tuning language models the next.

Most recently I spent two and a half years as a data scientist at the Virginia Retirement System, where I shipped NLP pipelines for compliance reporting, factor scoring models, and analytics consumed daily by the Equity, Risk, and Treasury teams. Before that I worked in energy markets on operational data feeding industrial price models, and I recently wrapped a computer-vision capstone at the UVA School of Medicine on surgical instrument recognition.

I have an MS in Data Science from the University of Virginia (4.0 GPA) and a quantitative undergrad in Math and Economics.

02 / Work

Selected experience.

2023 — 2025

Data Scientist

Virginia Retirement System  ·  Richmond, VA
  • Applied NLP and LLM-based sentiment analysis to 10-K/Q filings, news, and FOMC communications, producing macro signals consumed by the investment team for quarterly outlook reporting.
  • Built an OCR + NLP document-processing pipeline for SEC ADV compliance, cutting quarterly manual review by ~50% and improving audit traceability.
  • Engineered SQL stored procedures and views supporting Treasury and Risk reporting
  • Led cross-team process redesign and automation initiatives across Real Assets, Credit Strategies, Risk, Treasury, and external vendors (Snowflake, BarraOne), reducing end-to-end processing time by 30%+
  • Partnered with Equity, Risk, and Treasury stakeholders to ship Tableau, Streamlit, and Excel/VBA tools, including a cash-visibility dashboard that surfaced previously unrecognized free cash for the Treasury team.
  • Developed and maintained Python- and MATLAB-based factor scoring models and automated ETL pipelines, including exploratory data analysis on structured and document-based data such as PDF and XML, improving data integrity and reducing manual upkeep.
2022 (contract)

Data Analyst

Tabors Caramanis Rudkevich  ·  Energy markets consultancy
  • Processed 5M+ records of generator production data to derive operational parameters (min output, min up/down times, ramp constraints) feeding industrial price models and trading simulations.
  • Built scraping and ingestion pipelines across 10+ facilities, adding 200k+ historical data points and correcting previously misreported entries.
  • Developed anomaly detection logic to automatically flag forced outages and equipment downtime, reducing manual review.
2018 — 2022

Earlier analytics work

Various organizations  ·  Part-time & project-based
  • Time-series forecasting (ARIMA), anomaly detection, and visualization on operational datasets including municipal waste collection and athletic-performance data, supporting investigative analysis and policy evaluation.
03 / Projects

Featured work.

Surgical instrument recognition

UVA School of Medicine · Capstone

Trained YOLO11 + OpenCV computer vision models to segment and classify 25+ surgical instruments from operating-room video, cross-referenced against surgeon preference cards to quantify utilization and identify waste-reduction opportunities. Delivered to clinical stakeholders at the PeriOp Green academic lab.

Source materials are not publicly distributed at the lab’s preference; high-level summary and public-dataset companion available on request.

Real-time YOLO11 instrument detection — capstone demo

LLM fine-tuning, published on Hugging Face

Personal · Open-source

Fine-tuned an open-weights 8B base model with PEFT/AdaLoRA for a headline classification. Published model card, evaluation methodology, and results on Hugging Face alongside reproducible training code.

04 / Skills

Tools of the trade.

Languages
Python (pandas, NumPy, scikit-learn, PyTorch, Keras), SQL, R, MATLAB, VBA
ML & AI
NLP (spaCy, NLTK, Hugging Face Transformers), LLM APIs (OpenAI, Anthropic), PEFT/LoRA fine-tuning, computer vision (YOLO, OpenCV), supervised & unsupervised learning, Bayesian modeling, time-series (ARIMA), regression
Data & Infra
Snowflake, ETL pipelines, SQL stored procedures, OCR, PDF/XML ingestion, data validation, web scraping, workflow automation (PowerAutomate), Git, Jira (agile)
Visualization & BI
Tableau, Streamlit, Plotly, Excel (PivotTables, VBA, VLOOKUP)
05 / Contact

Get in touch.

I’m actively looking for my next role — full-time data scientist, data analyst, or ML engineer positions, either remote or on-site (open to relocation). The fastest way to reach me is email or LinkedIn.

Email
LinkedIn /in/daniel-sery
Hugging Face /dms3g
GitHub add link

Résumé

Download a copy of my résumé as a PDF.