James Kendrick
Software Engineer · Data Engineer · Systems/Platform · Backend
Summary
I am currently seeking 2027 new-grad opportunities in software engineering.
I gravitate toward software engineering at the backend, platform, and infrastructure layers that other engineers build on top of. What excites me most is designing scalable, intuitive developer tooling and novel abstractions that compress complexity into clean, composable APIs.
At Meta, this meant building a configuration-driven pipeline generator that let Data Scientists onboard new metric funnels in minutes instead of days, the goal being to encode domain expertise into an abstraction expressive enough that boilerplate disappears. That same instinct shows up in tools like Sprite, where I mapped navigation behavior into a small, self-contained object model that learns from you over time. I think deeply about object-oriented design, not as a methodology to follow, but as a way of finding the natural seams in a problem and giving them first-class representation in code. The best, most satisfying systems I've worked on feel inevitable in hindsight, where each piece is extensible, each interface is obvious, and each abstraction earns its value.
Education
University of Illinois Urbana-Champaign
B.S. Mathematics & Computer Science
GPA: 3.86 / 4.00
Aug 2023 – May 2027
Relevant Courses: Data Structures · Algorithms · Database Systems · Computer Systems · Programming Languages
Not-so-relevant Courses: Real Analysis · Nonlinear Programming · Abstract Linear Algebra · Models of Computation
Experience
Fab2· Software Engineer InternINCOMINGSep – Dec 2026
San Francisco, CA
- Developing data-centric applications and infrastructure for a software-defined semiconductor fab.
Meta· Data Engineer InternJun – Aug 2026
Seattle, WA
- Architected an artifact-agnostic drift detection framework that keeps AI-generated artifacts in sync with the specs they derive from; used a scalable design where each artifact registers its logic behind a shared core that orchestrates execution, reporting, and logging.
- Implemented adapters for two artifact classes covering 36 items and 12 specs, unmarshalling spec text, Thrift types, and iOS/Android source code into canonical, diffable models; surfaced 20+ actionable drift items on first run.
- Built a Python library and agent skill letting AI tools auto-remediate drift, backed by a new server-side endpoint written in PHP that dropped legacy dev server dependencies, which cut a single edit from 5+ minutes and a manual 2FA confirmation to ~10 seconds.
- Shipped the messaging observability data layer for Meta’s standalone Seller Marketplace app: owned pipelines computing health metrics for 4 user-facing flows, plus DQ validation, alerting wiring, and a leadership-facing dashboard.
Meta· Data Engineer InternMay – Aug 2025
New York City, NY
- Designed a configuration-driven Python ETL framework that horizontally scaled dimensional modeling across the IG Graph domain, enabling others to define arbitrary event chains and deploy multi-terabyte pipelines in minutes.
- Optimized the framework’s procedural SQL translator to execute a disk-backed, Breadth-First Search (BFS) data cube lattice, bypassing Presto out-of-memory limitations and slashing peak memory utilization by ~80% across all generated pipelines.
- Designed and deployed large-scale data pipelines in Python, powering interactive dashboards that analyze and monitor the user experience of discovering and connecting with friends on Instagram, partnering with DSs and DEs.
- Improved private follow surface attribution accuracy by 4% in existing pipelines and overall surface attribution by 20% in new ones, impacting 30+ downstream metrics consumed by 100+ users and enabling more reliable product decision-making.
- Orchestrated complex dependency graphs within Dataswarm (Meta’s internal Airflow), enforcing strict SLAs for 30+ downstream systems consumed by 100+ internal stakeholders.
Projects
Click a project to expand details →
A meta-runtime for orchestrating polyglot computation, built in Go. Define functions in Python or Go (the runtime-adapter interface is pluggable, so any language can be added), describe them in a spell.yaml or a custom DSL — or let Grimoire infer the signature from the source itself — and Grimoire generates fully typed CLI commands, handling argument parsing, type coercion, and runtime dependency provisioning (Python venvs, Go binaries) automatically. Uses Tree-sitter for AST-based function signature extraction, eliminating manual config.
A production-grade microservices platform for fantasy basketball managers. Engineered a Genetic Algorithm in Go that solves the NP-Hard lineup optimization problem — evaluating ~10⁹ combinations via concurrent population evolution, consistently outperforming greedy heuristics by ~15% in projected fantasy points.
A Zsh augmentation tool built in C++20 that learns from your navigation patterns. Embeds SQLite for persistent history and uses a frequency/recency ranking algorithm to surface the right directory instantly — reducing keystrokes by ~40% during deep traversal.
A containerized application that plugs into any PostgreSQL or MySQL database and introspects schemas at runtime, constructing a directed graph of foreign key relationships. Uses BFS to find the shortest JOIN path between any two tables — automatically generating valid multi-table queries without manual input.
A published Rust library (crates.io v0.1.2) for building personalized recommendation systems. Uses Candle to run pre-trained transformer models locally, generating vector embeddings for cosine similarity clustering and tag-based filtering. This isn't my proudest project, but it's honest work!
Technical Skills
About Me
I'm from San Antonio, Texas, and grew up loving mathematics. My love for math originally stemmed from the deep satisfaction of building an intuitive understanding of a concept and then mapping a complex problem onto it. This led me to start off my academic career as a math major, but when I discovered programming during my first year at UIUC, I realized the problem-solving process was identical, but with an incredible addition: creative freedom. Instead of being bound by existing mathematical rules, object-oriented programming allows me to invent my own constructs. CS took the analytical reasoning I loved in math and turned it into a sort of art form, allowing me to design elegant, custom tools that perfectly mirror my thought process.
When I'm not coding, I like to stay active. I played baseball my entire life through high school, so to scratch that athletic itch, I've now channeled that energy more diffusely into weightlifting, tennis, disc golf, basketball, and backpacking. Of those, I think going on long, overnight backpacking trips and enjoying nature with friends is my favorite. Packing up a week's worth of food and gear and then heading out into the Yosemite wilderness to scramble up some sketchy peaks or take a dip in some frigid alpine lakes is my idea of a perfect, relaxing vacation. I also love playing co-op video games with friends, finding new music to listen to, spending time with my family and friends, and recently, photography (especially landscapes on my backpacking trips). Oh, and if it wasn't obvious by my Court Vision project and being a San Antonio native, I'm a huge fantasy basketball and Spurs fan — Go Spurs Go!
Prefer a PDF? jlkendrick.dev/resume.pdf
click to expand