Projects
Embedded Video Preprocessing Pipeline — AMD Xilinx Kria FPGA (NDA Protected)
This project involves developing an embedded video preprocessing and validation pipeline deployed on an AMD Xilinx Kria FPGA platform. The system is designed to support downstream computer vision and machine learning workflows by ensuring that only valid, trustworthy input data enters later stages of processing.
The work emphasizes real-time operation, deterministic behavior, and reliability under resource constraints; concerns that closely mirror those found in safety-critical and regulated environments, including medical technology.
Due to the nature of the project, implementation details, performance characteristics, and system internals are protected under a non-disclosure agreement. At a high level, the project focuses on:
- Embedded Linux development on heterogeneous hardware
- Reasoning about data integrity, failure modes, and system reliability
- Designing software that prioritizes correctness and predictable behavior
- Operating within tight hardware, latency, and resource constraints
I’m happy to discuss the system architecture, design tradeoffs, and general engineering approach at a high level in interview settings, while fully respecting NDA obligations.
Current focus
- Refining data validity metrics and thresholds for real-time decision-making
- Improving robustness and observability of the preprocessing pipeline
- Evaluating tradeoffs between latency, accuracy, and resource usage on embedded hardware
JobApp — Structured Job Application Tracker (CLI + SQLite)
JobApp is a lightweight command-line tool for tracking job and internship applications using a real database schema instead of ad-hoc spreadsheets. I built it to support a structured, automation-friendly workflow and to demonstrate how quickly I can design and ship a usable piece of software.
The tool uses SQLite as a backing store and exposes a simple, expressive CLI for managing applications throughout the hiring pipeline.
What it does
- Records applications with company, role, location, source, status, priority, and notes
- Tracks application state across common pipeline stages (Applied, Recruiter Screen, OA, Interview, Offer, etc.)
- Makes it easy to identify stalled applications that may need follow-up
- Exports data to CSV for analysis or reporting in other tools
Automation and AI integration
The repository includes a CONVENTIONS.md file that defines a strict command format and vocabulary for interacting with the CLI. I use this document as context for large language models, allowing me to paste in a job description and reliably generate a structured jobapp add ... command that inserts consistent records into the database.
This approach combines deterministic tooling with LLM-assisted input, while keeping the source of truth fully local and auditable.
Tech stack / skills demonstrated
- Python 3 CLI design using
argparse - Relational schema design and querying with SQLite
- Clear separation between CLI, database access, and domain logic
- Modern Python packaging with
pyproject.tomland console entry points - Automated testing with
pytest - GitHub Actions CI for continuous test validation
Source