Apr 2026 — Present
Twinloop · Santa Clara, CA
Senior Software Engineer
Building some useful AI agents for marketing and brands.
Systems Engineer · AI Security · Seattle, Washington
🙏 Namaste,
I’m TJ
I build systems software and work on AI security where latency, reliability, and real-world constraints matter.
My work spans real-time systems, robotics, and the security of AI systems — from agent and model safety to
resilient infrastructure underneath them.
I’ve worked on teleoperation over spotty LTE/4G networks,
event-driven sensing for autonomous systems, and production backends that had to stay dependable amid
scale, noise, and unexpected failures.
I work across C++, Python, and cloud infrastructure, always prioritizing observability, resilience, and
top-notch performance.
I love creating systems that remain calm and collected, no matter how
chaotic the world around them gets — including when the chaos comes from the AI itself.
Systems engineering
AI security
Real-time systems
Robotics software
Observability
<200 ms
Telemetry and sensing pipelines under real-world timing constraints
p99 ~5s → <1s
Latency reduction in production systems workflows
Millions/day
Transactions and events handled in production systems
Experience
I’ve built across startups, research, and production systems.
Different settings, same pull: software that had to perform, stay reliable, and remain debuggable under pressure.
Sep 2025 — Jan 2026
mundane · Palo Alto, CA
Systems Software Engineer
Built a correctness-first, low-latency teleoperation data plane in C++ for video and telemetry over unreliable networks, maintaining sub-200ms tail latency under live bandwidth variability.
- Built packetization, sequencing, buffering, and failure-handling mechanisms for real-time streams.
- Added end-to-end instrumentation for latency, jitter, and packet loss.
- Deployed and operated latency-sensitive services on GCP with Docker, owning observability, debugging, and production reliability end to end.
Oct 2024 — Dec 2025
George Washington University · Washington, D.C.
R&D Engineer
Built real-time, event-driven autonomy pipelines for drones and rovers, integrating sensing, control, and fault-detection systems under strict latency and reliability constraints.
- Modified PX4-based flight-control software in C++ through custom messaging, sensor integration, and adversarial test workflows.
- Architected sub-200ms sensor and coordination pipelines over UDP for anomaly detection and recovery.
- Built simulation and replay tooling for fault injection and diagnosis, surfacing integration bugs that had blocked field testing.
Jan 2025 — May 2025
George Washington University · Washington, D.C.
Teaching Assistant
Supported a graduate and undergraduate course on autonomous systems, helping teach the fundamentals of controls, PID tuning, and autonomous vehicle design through hands-on rover development.
- Taught the basics of feedback control, PID behavior, and system tuning for real robotic platforms.
- Helped students design, build, and debug an autonomous rover across sensing, control, and integration workflows.
- Worked closely with students on implementation details, failure cases, and the gap between clean theory and messy physical systems.
Jan 2022 — Jan 2024
Gale Partners · New York, NY
Backend Developer
Built and fixed production backend systems in a high-pressure financial environment, with a focus on latency, reliability, and operational clarity.
- Eliminated a systemic p95 latency regression in core banking APIs, reducing latency from 5s to under 1s.
- Built event-driven transaction pipelines for millions of daily events with robust failure handling and reliable consumer behavior.
- Shipped shared middleware for auth, tracing, and rate limiting across six microservices.
Jan 2020 — Jan 2022
Follo Platforms · Bengaluru, India
Founding Engineer
Built the initial product and platform from the ground up, making core product and technical decisions without an existing playbook.
- Shipped customer-facing features every week for two years.
- Built a real-time rules engine that routed alerts within 500ms, reducing missed critical incidents by about 40%.
- Owned all production infrastructure, autoscaling, monitoring, deployment, and runbooks as the sole infra owner, maintaining 99.9% uptime.
Projects
Projects grounded in real systems problems.
A few things I’ve built around observability, reliability, robotics, simulation, and software that has to behave well under real constraints.
Vault Physician: Obsidian Plugin
AI-powered vault health monitor for Obsidian. Finds orphaned notes, stale content, tag sprawl, and broken structure with one-click fixes.
Obsidian
AI
Knowledge Management
Flint
A session firewall for MCP-connected agents. Enforces RBAC - roles, bindings, scopes, and per-argument constraints - then tracks data flow across calls to catch exfiltration attacks that per-request inspection misses. Selected for Canopy by Founders Inc (f.inc).
MCP
AI Security
Agents
Companion Health
A low-level C++ MAVLink health daemon for companion computers, designed to expose hardware and service health under fault conditions through a custom health message protocol.
C++
MAVLink
Health Monitoring
Drone Builder
A collaborative drone design and simulation platform that combines real-time 3D visualization with deterministic validation of configurations and flight setups.
3D Visualization
Simulation
Drone Design
Early Mission Degradation Detection in Multi-Agent Autonomous Systems
Built a multi-agent reinforcement learning project (ExoRL) involving environment modeling, agent training, and
experimental evaluation. Explored coordination strategies and analyzed emergent behaviors through simulation
results.
Advanced ML course, Fall '25 - Prof. Armin Mehrabian & Sardar Hamidian
Reinforcement Learning
Multi-Agent Systems
Simulation
Python
Sysmon
A real-time system monitoring tool with a Go backend and D3.js-based frontend visualizations. It monitors CPU, memory, disk, and network usage.
Go
D3.js
Monitoring
Homomorphic Encryption Toolkit
A toolkit built with C++ and Go to experiment with encrypted data operations for privacy-preserving computation and research.
C++
Go
Homomorphic Encryption
Lunar Lander - Hand Gesture Controlled Game
Control the OpenAI Gym LunarLander-v2 environment using hand gestures detected from a webcam with MediaPipe Hands and OpenCV.
Python
OpenCV
MediaPipe
OpenAI Gym
Research
My research sits at the intersection of autonomy, resilience, and security.
I’m especially interested in autonomous systems that have to keep working under latency, packet loss, adversarial conditions, and real-world uncertainty and in the cryptographic systems that make trust possible in distributed environments.
Autonomous Systems & Robotics
Low-latency autonomy under adversarial conditions
I work as a Research Engineer with Prof. Sibin Mohan at George Washington University, focusing on autonomous drones and rovers with an emphasis on secure, low-latency communication and system resilience.
- Real-time communication under packet loss and latency
- Multi-agent coordination and telemetry
- Security in adversarial environments
ROS
PX4
UDP Pipelines
Vicon
Jetson
Raspberry Pi
Current Focus
Failure detection, protocol weakness, and mitigation
My current work looks at how autonomous systems degrade before failure becomes obvious - from memory-safety issues in drone communication stacks to protocol-level weaknesses in onboard and ground-station interfaces, and real-time detection and mitigation of adversarial behavior.
The goal is to improve the safety, robustness, and reliability of autonomous systems operating in hostile or unpredictable environments.
Blockchain & Cryptography
Privacy-preserving and decentralized systems
I also have a research background in cryptography and decentralized systems, with a focus on privacy-preserving technologies such as zero-knowledge proofs.
- Elliptic curve cryptography
- Smart contract vulnerability analysis
- Privacy-aware system design
Writing and teaching technical ideas clearly
I co-authored Blockchain for Beginners: Art of Decentralization and Cryptography, a foundational text on cryptographic primitives and decentralized systems.
I’ve also delivered talks and workshops on blockchain and cryptography at conferences and technical meetups.
Skills
Strongest where systems meet software reality.
Languages & services
- C++
- Python
- Go · SQL
- FastAPI · Flask · Django · gRPC
- PostgreSQL · MongoDB · MySQL · Redis
Systems & infrastructure
- UDP and real-time networking
- Docker · GCP · AWS
- Kafka · RabbitMQ · Redis Streams
- CI/CD · autoscaling · observability · SLOs
Robotics & applied research
- PX4 · MAVLink · Vicon · RealSense
- ROS and autonomy pipelines
- Multi-agent systems · real-time inference
- Anomaly detection · RL · simulation
AI & agent security
- MCP session firewalls · agent sandboxing
- RBAC for agents · scopes · per-argument policy
- Protocol-level weakness analysis
- Adversarial behavior detection & mitigation
Beyond work
A few things about me.
The work matters, but so do the things that shape how I think, what I notice, and what I end up obsessing over.
Curiosity
Formula 1, systems, and engineering rabbit holes
I’m drawn to things that look elegant from a distance and get complicated up close - Formula 1, control under constraint, aircraft geometry, and the engineering decisions behind systems that somehow work as well as they do.
I write poetry, and I’m learning Farsi. Language sharpens the same instincts I use in engineering: structure, ambiguity, rhythm, and precision.
Thinking
Best thinking happens in motion
I like long hikes, road trips, and karting. Nature and speed usually help me untangle problems that refuse to move at a desk.
Art
Digital art and sketching
I sketch digitally when I want to think with my hands instead of code. It’s a different kind of problem-solving - slower, quieter, and a good reminder that not everything interesting needs to be optimized.
Books
Reading outside the screen
I like books that make me think differently about systems, people, and perception. Reading helps me step out of the immediacy of engineering and come back with better judgment.
I’m open to roles where the work is thoughtful, difficult, and real.
Systems and AI security roles are the best fit. I’m especially interested in teams working on
AI safety and security, real-time systems,
resilient infrastructure, autonomy, observability, or software that has to operate under real constraints.
If that sounds close to what you’re building, I’d be glad to talk.