Follow me on GitHub

Download resume (PDF)

Sepehr Laal's CV

Education

Portland State University, MSc. in Signal Processing -- Portland, OR

MSc.

2019 – 2023

Portland State University, BSc. in Electrical Engineering -- Portland, OR

BSc.

2016 – 2018

Portland Community College, ASc. in General Engineering -- Portland, OR

ASc.

2013 – 2015

Publications

BodySLAM: Feasibility of Tracking Human Kinematics with Simultaneous Localization and Mapping (SLAM)

Apr 2023

Thesis version of Sensors (2022) article; advisor James McNames

Sepehr Laal

10.15760/etd.3557 (M.S. Thesis, Portland State University, ECE)

Feasibility of Tracking Human Kinematics with Simultaneous Localization and Mapping

Dec 2022

Journal article from thesis research (Special Issue on Smart and Personalized Healthcare); first author; PSU ECE

Sepehr Laal, Paul Vasilyev, Sean Pearson, Mateo Aboy, James McNames

10.3390/s22239378 (Sensors 2022, 22(23), 9378)

Validation of Media Using Fingerprinting

Sept 2021

Granted patent for media validation using wavelet decomposition fingerprinting; sole inventor; AWS Elemental

Sepehr Laal

patents.google.com/patent/US11134279B1/en (US 11,134,279 B1, Amazon Technologies Inc)

Honors and Awards

Experience

Citi Bank, Senior VP of AI/ML Engineering and RL Infrastructure -- Portland, OR

2025 – present

Citi Bank, Senior VP of Cloud Security and Infrastructure -- Portland, OR

2020 – 2025

Monada Inc., Programming Language Architect -- Tel Aviv, Israel

2022 – 2023

Stelligent Systems LLC, DevOps Automation Engineer II -- Portland, OR

2019 – 2020

RootStream, Co-Founder and CTO -- San Francisco, CA

2018 – 2019

Virtulabs LLC., Software Consulting and Contracting -- San Francisco, CA

2017 – 2019

AWS Elemental Technologies, Software Development Engineer II -- Portland, OR

2017 – 2018

Helio Interactive PDX, Software Engineer -- Portland, OR

2015 – 2017

Downstream PDX, Junior Software Engineer -- Portland, OR

2014 – 2015

Open Source Projects

Sparse Dynamics

2026

Antivirus and anti prompt injection tooling for AI harnesses - Guardian CLI scans agent skills before execution; Trypanophobe reference backend for policy and enforcement; Paladin (kernel-level harness safety) in development

LibSLAM

2018

Heavily modified ORB-SLAM2 packaged as a reusable Linux library; foundational visual-SLAM infrastructure for the BodySLAM human-kinematics research

Cinder Doppler

2015

Implementation of Microsoft Research's SoundWave technique for motion tracking from Doppler shifts in reflected audio

Cinder Wavelets

2016

Real-time wavelet decomposition in Cinder with side-by-side FFT visualization; early signal-processing work related to later audio-fingerprinting research

LiveCam

2016

Open-source high performance video streaming system built with GStreamer and Node.js

Tree Sitter EventRule

2023

Tree-sitter grammar for AWS EventBridge event patterns; ships rule2rego, a compiler that transpiles Event Rules to OPA Rego policies on the fly

Particle Filter Localization

2017

CUDA particle-filter library for real-time camera pose estimation on checkerboard scenes without PnP; MATLAB orchestration, adapted from JEI 2014 camera-tracking research - written from scratch with custom CUDA kernels

Invited Talks

Certifications

Skills

Primary Languages: ANSI C / C++, JavaScript, TypeScript, Rust

Other Languages: Python, PHP, GLSL, HLSL, Cg, C#, Shell, Matlab

Tooling & Libraries: Unity3D, UE4, ASP.NET, React, React Native, Expo SDK, CUDA, MQTT, RabbitMQ, WebSockets, OpenGL, Terraform, GStreamer, FFmpeg, DroneCI, CircleCI, Jenkins, Git, GitHub

Containers & Sandboxing: Docker, ECS, ECR, WASM, Web Workers, Landlocks

Databases: MongoDB, DocumentDB, MySQL, MSSQL, SQLite, PostgreSQL, Redis

Operating Systems: Ubuntu, Amazon Linux, CentOS, Red Hat Enterprise Linux, Windows

Cloud Infrastructure: CloudFormation, AWS CDK, Azure, GCP, Alibaba, Terraform, Pulumi

AI Harness & RL Infrastructure: LiteLLM, n8n, OpenCode, OpenChamber, Web Fetch, Markdown conversion, DeepEval, Prompt Caching, Cloakbrowser, RLHF, RL, Lakera

AI Local Inference: Ollama, vLLM, Qwen family, Gemma 4 architecture, LLamaFile, Llama.cpp, HF transformers