Colin Murphy

New York, NY · clnjmurph@gmail.com

Founding engineer and full-stack builder with 8+ years shipping end-to-end, AI-powered products in fast-moving, data-intensive environments. Strongest in applied AI — LLM services, evaluation, observability, and grounding for domain-specific accuracy — and comfortable owning the full stack around it, from Python backends to React/Next.js frontends and cloud infrastructure (AWS & GCP). I have built solo from zero-to-MVP and led a small, craft-focused engineering team.

Schedule a Chat LinkedIn

Experience

Full-Stack AI Engineer (Contract)

PACT

Mar 2026 - Jun 2026

Owned and extended a production LLM platform that matches patients to clinical trials, spanning a Next.js frontend and two Python backend services.

Extended the clinical-trial eligibility pipeline with new LangGraph processing steps and Langfuse traceability; built criteria-level evaluation harnesses and resolved structured-output failures to raise matching reliability.
Shipped patient-facing product features end to end — Next.js / React / TypeScript frontend through gRPC + FastAPI Python services to the LLM matching engine and EHR data layer.
Designed a three-layer env-var safety system (typed schema → pre-deploy CI lint → boot-time validation) and standardized CI across four repos with a blocking test gate and Playwright post-deploy smoke tests.
Collapsed a fragmented three-repo local stack into a single Makefile-driven command.

Founding Engineer / Head of Engineering

HealthyMe AI

Oct 2023 - Feb 2026

Sole engineer scaling the platform from zero-to-MVP, then built and led a team of four delivering AI-driven clinical decision-support tools.

Developed LLM pipelines using RAG and LangChain, grounding model outputs in domain literature for accurate, well-supported answers in high-stakes clinical decision-making.
Architected and led the build of an LLM-assisted annotation platform (React/TypeScript, Node.js, PostgreSQL) with OAuth role-based review workflows that accelerated expert labeling throughput; personally owned the database schema and migrations.
Built ingestion pipelines to parse, OCR, and extract insight from 100,000+ unstructured PDF patient reports and 9M+ images into a structured knowledge base.
Trained a custom classification model spanning 60+ dermatological conditions, with metrics surpassing best-in-class numbers in peer-reviewed literature.

Data Scientist - Machine Learning Engineer

Cytiva

Jan 2022 - Jul 2023

Established model development lifecycle and the deployment scheme for at-scale machine learning applications.
Conducted data analysis as needed for New Product Development and Manufacturing Sustainability — ANOVAs, sampling plans, specification alignment, designed experiments as well as more advanced techniques like machine learning for predictive modelling and interactive data visualizations.
Developed data tools and data applications for use in production and R&D.
Conducted experimental design and hypothesis testing for product development and product sustainability.

Data Scientist

Johnson & Johnson

Jun 2020 - Dec 2021

Led interdisciplinary team of scientists to introduce data collection and exploration for 2 implant cleaning systems.
Built data processing pipeline for industrial manufacturing lines. Built random forest classification models to predict the presence of implants in 12 different chemical processing stages.
Designed LSTM model for time-series forecasting of solution chemistry — predicting downstream rinse water conductivity for process optimization/sustainability.
Parse through archived pdf documents to collect quality related information and build a relationship graph tool.

Scientist (Chemistry)

Johnson & Johnson

Aug 2016 - Jun 2020

Tasked with managing the Residual Manufacturing Materials Management (RM3) program for all U.S. based manufacturing locations.
Manage and coordinate medical device validation testing projects with both internal and external customers.
Analyze data and prepare test reports for established Quality Assurance programs.
Perform investigate lab testing of medical devices to evaluate cleanliness.

Undergraduate Researcher

Temple University

Jan 2014 - May 2016

Projects:

Patterned Arrays of Gold Nanoframes Using High Resolution E-Beam Lithography (Renewable Energy Lab, College of Engineering)
Optical detection of hot electron induced dissociation of H2 on gold nanoparticles (Borguet Research Group, Department of Chemistry)

Skills

Applied AI & LLMs

LLMs RAG Fine-tuning Evaluation & Observability Grounding LangChain LangGraph Langfuse Claude Code MCP Servers

Languages, Full-Stack & Cloud

Python TypeScript React Next.js FastAPI gRPC Docker AWS GCP PostgreSQL Git

Data Science & Machine Learning

Scikit-learn Keras / TensorFlow PyTorch Pandas Plotly Seaborn / Matplotlib Tableau Gephi

Documentation & Communication

Markdown LaTeX Overleaf HTML

Education

Master of Science — Data Science

Drexel University

Mar 2019 - Jun 2022

GPA: 3.95

Bachelor of Science — Chemistry

Temple University

Aug 2012 - May 2016

GPA: 3.70 · Phi Beta Kappa

Publications & Project Papers

Publications

NewsTweet: A Dataset of Social Media Embedding in Online Journalism. Mujib, M. I., Heidenreich, H. S., Murphy, C. J., Santia, G. C., Zelenkauskaite, A., & Williams J. R. Arxiv Preprint (2020).

Plastically deformed Cu-based alloys as high-performance catalysts for the reduction of 4-nitrophenol. Menumerov, E., Gilroy, K. D., Hajfathalian, M., Murphy, C. J., McKenzie, E. R., Hughes, R. A., & Neretina, S. Catalysis Science & Technology 6, no. 14 (2016): 5737-5745.

Project Papers

E-Weaver: Sustainable Clothing Aggregation and Recommendation System (Part 1: Data acquisition and EDA) — paper available upon request
E-Weaver: Sustainable Clothing Aggregation and Recommendation System (Part 2: Modeling) (Colin Murphy, Matthew Merenich, Dayun Piao, Zifeng Wang) — paper available upon request
Deep Learning and Chemometrics: Quantitative and Qualitative Spectroscopy Interpretation of Aqueous Solutions (Colin Murphy)
The Cryptocurrency Filabuster: an Analysis of Blindly Signed Contracts as a Feather Forking Countermeasure (Colin Murphy)
Classifying Short Text Reddit Posts (Colin Murphy, Mithila Guha, & Angela Mastrianni)
Utilizing Machine Learning to Classify Chronic Disease: Quantifying Biopsychosocial Risk and Resilience (Colin Murphy, Meghan Colosimo, John Obuch, & Fatih Catpinar)
Impact of the COVID-19 Pandemic on BIPOC populations (Colin Murphy, Madhumitha Santhana Krishnan, Nicholas Lawrence, Joshua Lister, & Rebecca Topper)
Patterned Arrays of Gold Nanoframes Using High Resolution E-Beam Lithography (Colin Murphy, Maryam Haifathalian, Carrigan Braun, & Svetlana Neretina)

Interests

Outside of the digital realm I enjoy staying active by running, but my true passion is for cycling. I have been an avid cycler for over 10 years and prefer it as a mode of transportation in the city. I have also participated in the American Cancer Society Philadelphia Bike-a-thon (fundraising event) for the past 10 years — biking 65-100 miles from Philadelphia to Atlantic City.

I also like to "geek out" over analog synthesizers and producing different types of ambient music by experimenting with different soundscapes.

Like many others of my generation, I am very conscious of our impact on our environment and strive to incorporate sustainability into every aspect of my personal life, while also advocating for a more sustainable society as a whole.

Awards & Certifications

Computational Data Science Certification — Drexel University

Mental Health First Aid Certification — National Council for Behavioral Health

Honorable Mention (top 5) — Temple University Undergraduate Research Symposium 2014 & 2015

Temple University — Shirley and Bernard Brown Scholarship — 2015