Ben Truong
ML DevOps & LLM Research
Hello
I am a ML engineer with DevOps experience and cross-disciplinary background in NLP, public-sector technology, and operational analytics. I’m currently pursuing my M.S. in Computer Science (ML track) at Georgia Tech, while also leading ML analytics products and GenAI application pipelines in civic tech and nonprofit spaces.
My recent work includes a ML-driven dashboard for financial insights on K8s, a RL model for stock trading, and a scalable LLM evaluation pipelines for hospital coding. I bring over three years of leadership experience in enterprise operations and 8 years in military Signal operations. My focus is on reproducibility and documentation for joint development between business and technical teams.
Outside of work, I’m passionate about open-source tools, community mentoring, and ML on consumer hardware.
Download my resumé as a PDF.
- ML Analytics Pipeline
- NLP on Federated Learning
- Entrepreneurship & Nonprofits
M.S. Computer Science (Machine Learning), 2025
Georgia Institute of Technology
B.A. Cognitive Science, 2019
UC Berkeley
Experience
- Selected as shift leader and SME on internal ML tools for 2.5 years; coordinated cross-team data overrides.
- Built high-throughput KPI dashboard (30k req/sec) with <200ms latency using AWS CloudWatch.
- Optimized staffing with ML + SQL; improved fulfillment output by 12% sitewide.
- Led process improvement team achieving 98% YoY reduction in GiftWrap misses using Six Sigma methods.
- Led 12-person dev team across Django/Python/React stack with AWS Lambda + Heroku deployment.
- Founded PPE web app that distributed 140,000 units nationally during COVID-19.
- Raised $28k and collaborated with clinics and CDC stakeholders for equitable distribution.
- Deployed tactical networks over satellite and configured switches/routers per U.S. ComSec standards.
- Built secure distributed compute setups using VPN, SSH, and virtualization tools in live environments.
Education
- Fine-tuned RL with DynaQ models for stock trading, achieving a 3% gain over S&P 500 in back-tests.
- Improved ICU diagnosis model (Gangavarapu et al, 2020) from 80% to 98% accuracy using NLP on unstructured clinical notes.
- Built A* and Dijkstra path planning models with real-time Bayesian updates for simulated robot and network traffic environments.
- Used AIP platform to build an ETL pipeline and LLM evaluation dashboard for rapid model analysis and selection.
- GPA: 3.8 / 4.0
- Glushko Prize for thesis on Outgroup Alienation and Empathy Suppression: behavioral + neurological mechanisms.