Software Engineer @ Bloomberg
Hi, I'mUchenna Njoku
Driven by curiosity and the challenge of solving problems that matter.
3rd of 2,000+ in the Jamaica National Mathematical Olympiad. Selected to represent Jamaica at the Central American & Caribbean Mathematical Olympiad.
Solved problems at
Experience
BloombergCurrent
Software Engineer
Aug 2026 – Present
Goldman Sachs
Software Engineer Intern
Jun 2025 – Aug 2025
- Developed a secure, scalable CSV report generation and download system for the Payroll Tax Calculator using Spring Boot, AWS ECS Fargate, Lambda, S3 and Aurora PostgreSQL, with dynamic Java-based formatting and zero-knowledge password encryption — cutting processing time by 60% and meeting internal audit requirements.
- Led full-stack implementation of a U.S. Tax Reciprocity Rules Engine, designing OpenAPI-driven REST APIs and an accessible React + TypeScript UI for managing multi-state income tax exemptions — eliminating 95% of manual entries and reducing compliance errors to near zero, covering 46,000 employees in 30+ countries.
Cisco · Splunk
Software Engineer Intern
Jan 2025 – Apr 2025
- Engineered core functionality of Splunk's Maintenance Window Service in Go, automating ingestion of maintenance schedules from Google Sheets into Splunk — processing 10,000+ rows of operational data per week and reducing manual scheduling effort by 70%.
- Implemented checksum-based diffing and conflict-detection algorithms in Go, reducing redundant Splunk REST API calls by 30%, cutting duplicate schedule errors by 35%, and catching 98% of deployment conflicts (e.g. overlapping upgrade cohorts).
Amazon Web Services
Software Engineer Intern
May 2024 – Aug 2024
- Led the Local Compute Infrastructure (LCI) Bootloader Rollout project to streamline and automate deployment of network devices across the LCI fleet, impacting 500+ network sites globally.
- Built a release, rollout and deployment pipeline for new bootloader versions and PXE artifacts, with testing packages and environments — reducing bootloader release failures by 20%, expected to decrease system downtime by 30%.
Amazon Web Services
Software Engineer Intern
May 2023 – Aug 2023
- Architected the migration of the AWS Network host generation workflow to automated pipelines using CloudFormation, CloudWatch, Lambda and CDK, increasing site build speed by 40%.
- Increased robustness and manageability of 450+ AWS Network Centers running over 1 million external services, using TypeScript, Java and Python.
GM EcoCAR EV Challenge
Team Lead, Connected & Autonomous Vehicles
Jan 2023 – Dec 2024
- Led the Connected and Autonomous Vehicles (CAVs) sub-team, engineering vehicle control algorithms in MATLAB and Simulink and validating strategies in dSPACE and CARLA simulations.
- Refined large onboard sensor datasets with Fourier and wavelet transforms for noise reduction, using C++ for high-efficiency signal processing.
- Applied PCA and t-SNE for machine-learning-based vehicle health insights.
Research & featured project
Research
Outstanding Student Research Award, Florida Academy of Science
Emergent Communication in Multi-Agent Reinforcement Learning
Can agents that only see part of the world invent their own language to coordinate? I gave a team of reinforcement-learning agents a channel of a few discrete symbols, no meaning attached, and let training decide what to say.
Agent collisions
38.1%3.7%lower is better
baselinewith communication
Capture rate, held-out maps
65.8%89.5%higher is better
baselinewith communication
Channel entropy
3.99 / 4.00 bits
The agents use nearly every symbol: functional signalling, but not a symbolic language.
- Built a PettingZoo pursuit environment (16×16 grid, 3 pursuers vs. evasive prey, 7×7 partial observability) and trained shared-parameter DQN agents in PyTorch with a learned discrete communication channel (vocab 4 / 16), optimized end to end with no hand-coded protocol.
- Cut agent collisions from 38.1% to 3.7% and raised capture rate from 65.8% to 89.5% on held-out maps, isolating communication's value to out-of-distribution transfer.
- Quantified protocol structure via channel-entropy analysis (3.99 / 4.0 bits), showing functional but non-symbolic signaling.
- PyTorch
- PettingZoo
- DQN
- Python
- captures
- 0
- steps / capture
- —
- bump rate
- 0%
drag to rotate
- ATM vol, 1y
- 20.7%
- skew, 1y
- -0.347
- static arb
- none ✓
Featured project
VolSurf-CPP
Arbitrage-free volatility surface engine
A high-performance volatility surface engine in C++17 with Python bindings, built to take raw option quotes to a smooth, arbitrage-free implied volatility surface.
- Multi-threaded implied volatility solvers and constrained spline fitting, parallelised with OpenMP.
- Python bindings via pybind11, with pricers benchmarked against QuantLib.
- Automated research pipeline ingesting market data from Yahoo Finance and Polygon.io.
- Interactive 3D surface dashboards built with Plotly and Streamlit.
- C++17
- Python
- pybind11
- OpenMP
- QuantLib
Projects
WildTrack
NFC attendance system
Mobile attendance tracking using NFC for secure, contactless check-ins at educational institutions.
- React Native
- Supabase
- NFC
Background
Florida Academy of Science
Outstanding Student Research Award for work on emergent communication in multi-agent reinforcement learning.
Jamaica National Mathematical Olympiad
Placed 3rd of 2,000+ students; selected to represent Jamaica at the XVII Central American & Caribbean Mathematical Olympiad.
IEEE SoutheastCon Robotics
1st in Florida. Orchestrated hardware and software for a 4WD obstacle-navigating robot, deploying a Jetson Nano in a containerized Docker architecture.
Presidential Scholar
Bethune-Cookman University.
- Education
- B.S. Computer Science, Mathematics minor, Bethune-Cookman University, class of 2026 · Presidential Scholar
- Languages
- C++
- Python
- Java
- Go
- TypeScript
- JavaScript
- Rust
- C#
- Swift
- R
- SQL
- GraphQL
- Bash
- ML & data
- PyTorch
- TensorFlow
- NumPy
- pandas
- SciPy
- scikit-learn
- Systems & web
- AWS
- Docker
- Linux
- Spring
- React
- Node.js
- Angular
Contact
Let's talk.
Email is the best way to reach me.